RFC 6306






Internet Research Task Force (IRTF)                          P. Frejborg
Request for Comments: 6306                                     July 2011
Category: Experimental
ISSN: 2070-1721


                      Hierarchical IPv4 Framework

Abstract



   This document describes a framework for how the current IPv4 address
   space can be divided into two new address categories: a core address
   space (Area Locators, ALOCs) that is globally unique, and an edge
   address space (Endpoint Locators, ELOCs) that is regionally unique.
   In the future, the ELOC space will only be significant in a private
   network or in a service provider domain.  Therefore, a 32x32 bit
   addressing scheme and a hierarchical routing architecture are
   achieved.  The hierarchical IPv4 framework is backwards compatible
   with the current IPv4 Internet.

   This document also discusses a method for decoupling the location and
   identifier functions -- future applications can make use of the
   separation.  The framework requires extensions to the existing Domain
   Name System (DNS), the existing IPv4 stack of the endpoints,
   middleboxes, and routers in the Internet.  The framework can be
   implemented incrementally for endpoints, DNS, middleboxes, and
   routers.

Status of This Memo



   This document is not an Internet Standards Track specification; it is
   published for examination, experimental implementation, and
   evaluation.

   This document defines an Experimental Protocol for the Internet
   community.  This document is a product of the Internet Research Task
   Force (IRTF).  The IRTF publishes the results of Internet-related
   research and development activities.  These results might not be
   suitable for deployment.  This RFC represents the individual
   opinion(s) of one or more members of the Routing Research Group of
   the Internet Research Task Force (IRTF).  Documents approved for
   publication by the IRSG are not a candidate for any level of Internet
   Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6306.




Frejborg                      Experimental                      [Page 1]

RFC 6306               Hierarchical IPv4 Framework             July 2011


Copyright Notice



   Copyright (c) 2011 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.








































Frejborg                      Experimental                      [Page 2]

RFC 6306               Hierarchical IPv4 Framework             July 2011


Table of Contents



   1. Introduction ....................................................4
   2. Requirements Notation ...........................................7
   3. Definitions of Terms ............................................7
   4. Hierarchical Addressing .........................................9
   5. Intermediate Routing Architecture ..............................11
      5.1. Overview ..................................................11
      5.2. Life of a hIPv4 Session ...................................15
   6. Long-Term Routing Architecture .................................18
      6.1. Overview ..................................................19
      6.2. Exit, DFZ, and Approach Routing ...........................21
   7. Decoupling Location and Identification .........................23
   8. ALOC Use Cases .................................................24
   9. Mandatory Extensions ...........................................28
      9.1. Overview ..................................................28
      9.2. DNS Extensions ............................................29
      9.3. Extensions to the IPv4 Header .............................30
   10. Consequences ..................................................34
      10.1. Overlapping Local and Remote ELOC Prefixes/Ports .........34
      10.2. Large Encapsulated Packets ...............................35
      10.3. Affected Applications ....................................35
      10.4. ICMP .....................................................37
      10.5. Multicast ................................................37
   11. Traffic Engineering Considerations ............................38
      11.1. Valiant Load-Balancing ...................................39
   12. Mobility Considerations .......................................40
   13. Transition Considerations .....................................42
   14. Security Considerations .......................................43
   15. Conclusions ...................................................45
   16. References ....................................................47
      16.1. Normative References .....................................47
      16.2. Informative References ...................................47
   17. Acknowlegments ................................................50
   Appendix A. Short Term and Future IPv4 Address Allocation Policy ..51
   Appendix B. Multi-Homing becomes Multi-Pathing ....................53
   Appendix C. Incentives and Transition Arguments ...................57
   Appendix D. Integration with CES Architectures ....................58













Frejborg                      Experimental                      [Page 3]

RFC 6306               Hierarchical IPv4 Framework             July 2011


1.  Introduction



   A Locator/Identifier Separation Protocol [LISP] presentation from a
   breakout session at an expo held in January, 2008, triggered a
   research study; findings from the study are described in this
   document.  Further studies revealed that the routing community at
   IETF is concerned about the scalability of the routing and addressing
   system of the future Internet.  The Internet Architecture Board (IAB)
   held a Routing and Addressing workshop on October 18-19, 2006, in
   Amsterdam.  The outcome from the workshop is documented in [RFC4984].
   Also, the IRTF had established a Routing Research Group [RRG] in 2007
   and created some design guidelines; see [RFC6227].

   The author of this document found the LISP approach very interesting
   because the IP address space is proposed to be separated into two
   groups: Routing Locators (RLOCs), which are present in the global
   routing table of the Internet called the Default-Free Zone (DFZ), and
   Endpoint Identifiers (EIDs), which are only present in edge networks
   attached to the Internet.

   The proposed LISP architecture reduces the routing information in the
   DFZ, but it also introduces a new mapping system that would require a
   caching solution at the border routers installed between the edge
   networks and DFZ.  EID prefixes are not needed in the DFZ since a
   tunneling (overlay) scheme is applied between the border routers.  To
   the author, this seems to be a complex architecture that could be
   improved by applying lessons learned from similar past architectures
   -- in the '90s, overlay architectures were common, deployed on top of
   Frame Relay and ATM technologies.  Cache-based routing architectures
   have also been tried, for example, Ipsilon's IP Switching.  These
   architectures have largely been replaced by MPLS [RFC3031] for
   several reasons -- one being that overlay and caching solutions have
   historically suffered from scalability issues.  Technology has
   certainly evolved since the '90s.  The scalability issues of overlay
   and caching solutions may prove to be less relevant for modern
   hardware and new methods; see [Revisiting_Route_Caching]

   Nevertheless, the author has some doubt whether overlay and caching
   will scale well, based upon lessons learned from past overlay and
   caching architectures.  The hierarchical IPv4 framework proposal
   arose from the question of whether the edge and core IP addressing
   groupings from LISP could be used without creating an overlay
   solution by borrowing ideas from MPLS to develop a peer-to-peer
   architecture.  That is, instead of tunneling, why not swap IP
   addresses (hereafter called locators) on a node in the DFZ?  By
   introducing a shim header to the IPv4 header and Realm Border Router
   (RBR) functionality on the network, the edge locators are no longer
   needed in the routing table of DFZ.



Frejborg                      Experimental                      [Page 4]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Two architectural options existed regarding how to assemble the
   packet so that RBR functionality can be applied in the DFZ: the
   packet was assembled by either an ingress network node (similar to
   LISP or MPLS) or at the endpoint itself.  The major drawback in
   assembling the packet with a shim header at the endpoint is that the
   endpoint's stack must be upgraded; however, a significant advantage
   is that the Path MTU Discovery issue, as discussed in, e.g., LISP,
   would not exist.  In addition, the caching scalability issue is
   mitigated to the greatest extent possible by pushing caching to the
   endpoint.

   This approach also opened up the possibility of extending the current
   IP address scheme with a new dimension.  In an MPLS network,
   overlapping IP addresses are allowed since the forwarding plane is
   leveraging label information from the MPLS shim header.  By applying
   RBR functionality, extending the current IPv4 header with a shim
   header and assembling the new header at endpoints, an IP network can
   also carry packets with overlapping edge locators, although the core
   locators must still be globally unique.  The location of an endpoint
   is also no longer described by a single address space; it is
   described by a combination of an edge locator and a core locator, or
   a set of core locators.

   Later on, it was determined that the current 32-bit address scheme
   can be extended to 64 bits -- 32 bits reserved for globally unique
   core locators and 32 bits reserved for locally unique edge locators.

   The new 64-bit addressing scheme is backwards compatible with the
   currently deployed Internet addressing scheme.

   By making the architectural decisions described above, the foundation
   for the hierarchical IPv4 framework was laid out.

   Note that the hierarchical IPv4 framework is abbreviated as hIPv4,
   which is close to the abbreviation of Host Identity Protocol (HIP)
   [RFC4423].  Thus, the reader needs to pay attention to the use of the
   two abbreviations -- hIPv4 and HIP, which represent two different
   architectures.

   Use of the hIPv4 abbreviation has caused much confusion, but it was
   chosen for two reasons:

   o Hierarchical - to emphasize that a hierarchical addressing scheme
     is developed.  A formalized hierarchy is achieved in the routing
     architecture.  Some literature describes today's Internet as
     already using hierarchical addressing.  The author believes that
     this claim is not valid -- today's Internet uses one flat address
     space.



Frejborg                      Experimental                      [Page 5]

RFC 6306               Hierarchical IPv4 Framework             July 2011


     It is true that we have hierarchical routing in place.  A routing
     architecture can consist of at least three types of areas: stub
     area, backbone area, and autonomous system (AS).  The current flat
     address space is summarized or aggregated at border routers between
     the areas to suppress the size of a routing table.  In order to
     carry out summaries or aggregates of prefixes, the address space
     must be continuous over the areas.

     Thus, the author concludes that the current method is best
     described as an aggregating addressing scheme since there are
     address block dependencies between the areas.  Dividing addresses
     into edge and core locator spaces (a formalized hierarchy) opens up
     a new dimension -- the edge locator space can still be deployed as
     an aggregating address scheme on the three types of areas mentioned
     earlier.  In hIPv4, the core locators are combined with edge
     locators, independent from each other -- the two locator space
     allocation policies are separated and no dependencies exist between
     the two addressing schemes in the long-term architecture.

     A new hierarchical addressing scheme is achieved: a two-level
     addressing scheme describing how the endpoint is attached to the
     local network and also how the endpoint is attached to the
     Internet.  This change in the addressing scheme will enable a
     fourth level, called the Area Locator (ALOC) realm, at the routing
     architecture.

   o IPv4 - to emphasize that the framework is still based upon the IPv4
     addressing scheme, and is only an evolution from the currently
     deployed addressing scheme of the Internet.

   While performing this research study, the author reviewed a previous
   hierarchical addressing and routing architecture that had been
   proposed in the past, the Extended Internet Protocol (EIP) [RFC1385].
   Should the hIPv4 framework ever be developed from a research study to
   a standard RFC, it is recommended that the hierarchical IPv4
   framework name be replaced with Extended Internet Protocol, EIP,
   since both architectures share similarities, e.g., backwards
   compatibility with existing deployed architecture, hierarchical
   addressing, etc., and the hIPv4 abbreviation can be mixed up with
   HIP.

   This document is an individual contribution to the IRTF Routing
   Research Group (RRG); discussions between those on the mailing list
   of the group have influenced the framework.  The views in this
   document are considered controversial by the IRTF Routing Research
   Group (RRG), but the group reached a consensus that the document
   should still be published.  Since consensus was not achieved at RGG
   regarding which proposal should be preferred -- as stated in



Frejborg                      Experimental                      [Page 6]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   [RFC6115]: "The group explored a number of proposed solutions but did
   not reach consensus on a single best approach" -- thus, all proposals
   produced within RRG can be considered controversial.

2.  Requirements Notation



   The key words MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD,
   SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL in this document are to be
   interpreted as described in [RFC2119].

3.  Definitions of Terms



   This document makes use of the following terms:

   Regional Internet Registry (RIR):

      This is an organization overseeing the allocation and registration
      of Internet number resources within a particular region of the
      world.  Resources include IP addresses (both IPv4 and IPv6) and
      autonomous system numbers.

   Locator:

      A name for a point of attachment within the topology at a given
      layer.  Objects that change their point(s) of attachment will need
      to change their associated locator(s).

   Global Locator Block (GLB):

      An IPv4 address block that is globally unique.

   Area Locator (ALOC):

      An IPv4 address (/32) assigned to locate an ALOC realm in the
      Internet.  The ALOC is assigned by an RIR to a service provider.
      The ALOC is globally unique because it is allocated from the GLB.

   Endpoint Locator (ELOC):

      An IPv4 address assigned to locate an endpoint in a local network.
      The ELOC block is assigned by an RIR to a service provider or to
      an enterprise.  In the intermediate routing architecture, the ELOC
      block is only unique in a geographical region.  The final policy
      of uniqueness shall be defined by the RIRs.  In the long-term
      routing architecture, the ELOC block is no longer assigned by an
      RIR; it is only unique in the local ALOC realm.





Frejborg                      Experimental                      [Page 7]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   ALOC realm:

      An area in the Internet with at least one attached Realm Border
      Router (RBR).  Also, an ALOC must be assigned to the ALOC realm.
      The Routing Information Base (RIB) of an ALOC realm holds both
      local ELOC prefixes and global ALOC prefixes.  An ALOC realm
      exchanges only ALOC prefixes with other ALOC realms.

   Realm Border Router (RBR):

      A router or node that is able to identify and process the hIPv4
      header.  In the intermediate routing architecture, the RBR shall
      be able to produce a service, that is, to swap the prefixes in the
      IP header and locator header, and then forward the packet
      according to the value in the destination address field of the IP
      header.  In the long-term routing architecture, the RBR is not
      required to produce the swap service.  Instead, the RBR can make
      use of the Forwarding Indicator field in the locator header.  Once
      the FI-bits are processed, the RBR forwards the packet according
      to the value in the destination address of the IP header or
      according to the value in the ELOC field of the locator header.
      The RBR must have the ALOC assigned as its locator.

   Locator Header:

      A 4-byte or 12-byte field, inserted between the IP header and
      transport protocol header.  If an identifier/locator split scheme
      is used, the size of the locator header is further expanded.

   Identifier:

      The name of an object at a given layer.  Identifiers have no
      topological sensitivity and do not have to change, even if the
      object changes its point(s) of attachment within the network
      topology.

   Identifier/locator split scheme:

      Separate identifiers used by applications from locators that are
      used for routing.  The exchange of identifiers can occur
      discreetly between endpoints that have established a session, or
      the identifier/locator split can be mapped at a public database.









Frejborg                      Experimental                      [Page 8]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Session:

      An interactive information exchange between endpoints that is
      established at a certain time and torn down at a later time.

   Provider Independent Address Space (PI addresses/prefixes):

      An IPv4 address block that is assigned by a Regional Internet
      Registry directly to a user organization.

   Provider Aggregatable Address Space (PA addresses/prefixes):

      An IPv4 address block assigned by a Regional Internet Registry to
      an Internet Service Provider that can be aggregated into a single
      route advertisement.

   Site mobility:

      A site wishing to change its attachment point to the Internet
      without changing its IP address block.

   Endpoint mobility:

      An endpoint moves relatively rapidly between different networks,
      changing its IP layer network attachment point.

   Subflow:

      A flow of packets operating over an individual path, the flow
      forming part of a larger transport protocol connection.

4.  Hierarchical Addressing



   The current IP addressing (IPv4) and the future addressing (IPv6)
   schemes of the Internet are unidimensional by their nature.  This
   limitation -- the unidimensional addressing scheme -- has created
   some roadblocks, for example, breaking end-to-end connectivity due to
   NAT, limited deployment of Stream Control Transmission Protocol
   (SCTP) [RFC4960], etc., for further growth of the Internet.

   If we compare the Internet's current addressing schemes to other
   global addressing or location schemes, we notice that the other
   schemes use several levels in their structures.  For example, the
   postal system uses street address, city, and country to locate a
   destination.  To locate a geographical site, we use longitude and
   latitude in the cartography system.  The other global network, the
   Public Switched Telephone Network (PSTN), has been built upon a
   three-level numbering scheme that has enabled a hierarchical



Frejborg                      Experimental                      [Page 9]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   signaling architecture.  By expanding the current IPv4 addressing
   scheme from a single level to a two-level addressing structure, most
   of the issues discussed in [RFC4984] can be solved.  Also, a
   hierarchical addressing scheme would better describe the Internet we
   have in place today.

   Looking back, it seems that the architecture of the Internet changed
   quite radically from the intended architecture with the introduction
   of [RFC1918], which divides the hosts into three categories and the
   address space into two categories: globally unique and private
   address spaces.  This idea allowed for further growth of the Internet
   and extended the life of the IPv4 address space, and it ended up
   becoming much more successful than expected.  RFC 1918 didn't solve
   the multi-homing requirements for endpoints providing services for
   Internet users, that is, multi-homed sites with globally unique IP
   addresses at endpoints to be accessed from the Internet.

   Multi-homing has imposed some challenges for the routing architecture
   that [RRG] is addressing in [RFC6115].  Almost all proposals in the
   report suggest a core and edge locator separation or elimination to
   create a scalable routing architecture.  The core locator space can
   be viewed to be similar to the globally unique address space, and the
   edge locator space similar to the private address space in RFC 1918.

   RFC 1918 has already demonstrated that Internet scales better with
   the help of categorized address spaces, that is, globally unique and
   private address spaces.  The RRG proposals suggest that the Internet
   will be able to scale even further by introducing core and edge
   locators.  Why not then change the addressing scheme (both IPv4 and
   IPv6 addressing schemes, though this document is only focusing on
   IPv4) to better reflect the current and forthcoming Internet routing
   architecture?  If we continue to use a flat addressing scheme, and
   combine it with core (global) and edge (private) locator (address)
   categories, the routing architecture will have to support additional
   mechanisms, such as NAT, tunneling, or locator rewriting with the
   help of an identifier to overcome the mismatch.  The result will be
   that information is lost or hidden for the endpoints.  With a two-
   level addressing scheme, these additional mechanisms can be removed
   and core/edge locators can be used to create new routing and
   forwarding directives.

   A convenient way to understand the two-level addressing scheme of the
   hIPv4 framework is to compare it to the PSTN numbering scheme
   (E.164), which uses country codes, national destination codes, and
   subscriber numbers.  The Area Locator (ALOC) prefix in the hIPv4
   addressing scheme can be considered similar to the country code in
   PSTN; i.e., the ALOC prefix locates an area in the Internet called an
   ALOC realm.  The Endpoint Locator (ELOC) prefixes in hIPv4 can be



Frejborg                      Experimental                     [Page 10]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   compared to the subscriber numbers in PSTN -- the ELOC is regionally
   unique (in the future, locally unique) at the attached ALOC realm.
   The ELOC can also be attached simultaneously to several ALOC realms.

   By inserting the ALOC and ELOC elements as a shim header (similar to
   the MPLS and [RBridge] architectures) between the IPv4 header and the
   transport protocol header, a hIPv4 header is created.  From the
   network point of view, the hIPv4 header "looks and feels like" an
   IPv4 header, thus fulfilling some of the goals as outlined in EIP and
   in the early definition of [Nimrod].  The outcome is that the current
   forwarding plane does not need to be upgraded, though some minor
   changes are needed in the control plane (e.g., ICMP extensions).

5.  Intermediate Routing Architecture



   The intermediate routing architecture is backwards compatible with
   the currently deployed Internet; that is, the forwarding plane
   remains intact except that the control plane needs to be upgraded to
   support ICMP extensions.  The endpoint's stack needs to be upgraded,
   and middleboxes need to be upgraded or replaced.  In order to speed
   up the transition phase, middleboxes might be installed in front of
   endpoints so that their stack upgrade can be postponed; for further
   details, see Appendix D.

5.1.  Overview



   As mentioned in previous sections, the role of an Area Locator (ALOC)
   prefix is similar to a country code in PSTN; the ALOC prefix provides
   a location functionality of an area within an autonomous system (AS),
   or an area spanning over a group of ASes, in the Internet.  An area
   can have several ALOC prefixes assigned, e.g., for traffic
   engineering purposes such as load balancing among several
   ingress/egress points at the area.  The ALOC prefix is used for
   routing and forwarding purposes on the Internet, and so the ALOC
   prefix must be globally unique and is allocated from an IPv4 address
   block.  This globally unique IPv4 address block is called the Global
   Locator Block (GLB).

   When an area within an AS (or a group of ASes) is assigned an ALOC
   prefix, the area has the potential to become an ALOC realm.  In order
   to establish an ALOC realm, more elements, more than just the ALOC
   prefix, are needed.  One or multiple Realm Border Routers (RBRs) must
   be attached to the ALOC realm.  An RBR element is a node capable of
   swapping the prefixes of the IP header and the new shim header,
   called the locator header.  The swap service is described in detail
   in Section 5.2, step 3.





Frejborg                      Experimental                     [Page 11]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Today's routers do not support this RBR functionality.  Therefore,
   the new functionality will most likely be developed on an external
   device attached to a router belonging to the ALOC realm.  The
   external RBR might be a server with two interfaces attached to a
   router, the first interface configured with the prefix of the ALOC
   and the second with any IPv4 prefix.  The RBRs do not make use of
   dynamic routing protocols, so neither a Forwarding Information Base
   (FIB) nor a cache is needed -- the RBR performs a service, swapping
   headers.

   The swap service is applied on a per-packet basis, and the
   information needed to carry out the swap is included in the locator
   header of the hIPv4 packet.  Thus, a standalone device with
   sufficient computing and I/O resources to handle the incoming traffic
   can take the role as an RBR.  Later on, the RBR functionality might
   be integrated into the forwarding plane of a router.  It is expected
   that one RBR will not be able to handle all the incoming traffic
   designated for an ALOC realm and that having a single RBR would also
   create a potential single point of failure in the network.
   Therefore, several RBRs might be installed in the ALOC realm and the
   RBRs shall use the ALOC prefix as their locator, and the routers
   announce the ALOC prefix as an anycast locator within the local ALOC
   realm.  The ALOC prefix is advertised throughout the DFZ by BGP
   mechanisms.  The placement of the RBRs in the network will influence
   the ingress traffic to the ALOC realm.

   Since the forwarding paradigm of multicast packets is quite different
   from forwarding unicast packets, the multicast functionality will
   have an impact on the RBR.  Because the multicast RBR (mRBR)
   functionality is not available on today's routers, an external device
   is needed -- later on the functionality might be integrated into the
   routers.  The mRBR shall take the role of an anycast Rendezvous Point
   with the Multicast Source Discovery Protocol (MSDP) [RFC3618] and
   Protocol Independent Multicast (PIM) [RFC4601] capabilities, but to
   swap headers neither a FIB nor a cache is required.  As with the RBR,
   the multicast hIPv4 packets are carrying all needed information in
   their headers in order to apply the swap service; for details, see
   Section 10.5.

   The ALOC realm is not yet fully constructed.  We can now locate the
   ALOC realm on the Internet, but to locate the endpoints attached to
   the ALOC realm, a new element is needed: the Endpoint Locator (ELOC).
   As mentioned in the previous section, the ELOC prefixes can be
   considered similar to the subscriber numbers in PSTN.  The ELOC is
   not a new element but a redefinition of the current IPv4 address
   configured at an endpoint.  The term redefinition is applied because
   when the hIPv4 framework is fully implemented, the global uniqueness
   of the IPv4 addresses is no longer valid.  A more regional address



Frejborg                      Experimental                     [Page 12]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   allocation policy of IPv4 addresses can be deployed, as discussed in
   Appendix A.  The ELOC prefix will only be used for routing and
   forwarding purposes inside the local and remote ALOC realms, and it
   is not used in the intermediate ALOC realms.

   When an initiator is establishing a session to a responder residing
   outside the local ALOC realm, the value in the destination address
   field of the IP header of an outgoing packet is no longer the remote
   destination address (ELOC prefix); instead, the remote ALOC prefix is
   installed in the destination address field of the IP header.  Because
   the value in the destination address field of the IP header is
   carrying an ALOC prefix, the intermediate ALOC realms do not need to
   install the ELOC prefixes of other ALOC realms in their routing
   tables.  It is sufficient for the intermediate ALOC realms to carry
   only the ALOC prefixes.

   The outcome is that the routing tables at each ALOC realm will be
   reduced when the hIPV4 framework is fully implemented.  The ALOC
   prefixes are still globally unique and must be installed in the DFZ.
   Thus, the service provider cannot control the growth of the ALOC
   prefixes, but she/he can control the amount of local ELOC prefixes in
   her/his local ALOC realm.

   When the hIPv4 packet arrives at the remote ALOC realm, it is
   forwarded to the nearest RBR, since the value in the destination
   address field of the IP header is the remote ALOC prefix.  When the
   RBR has swapped the hIPv4 header, the value in the destination
   address field of the IP header is the remote ELOC; thus, the hIPv4
   packet will be forwarded to the final destination at the remote ALOC
   realm.  An endpoint using an ELOC prefix can be attached
   simultaneously to two different ALOC realms without the requirement
   to deploy a classical multi-homing solution; for details, see Section
   12 and Appendix B.

   Understanding that the addressing structure is no longer
   unidimensional and that a second level of hierarchy has been added,
   it is important to solve the problems of locating the remote ELOC
   (endpoint) and remote ALOC realm on the Internet, as well as
   determining where to assemble the header of the hIPv4 packet.  The
   hierarchical IPv4 framework relies upon the Domain Name System needs
   to support a new record type so that the ALOC information can be
   distributed to the endpoints.  To construct the header of the hIPv4
   packet, either the endpoint or an intermediate node (e.g., a proxy)
   should be used.  A proxy solution is likely to prove suboptimal due
   to a complication induced by the proxy's need to listen to DNS
   messages, and a cache solution has scalability issues.





Frejborg                      Experimental                     [Page 13]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   A better solution is to extend the current IPv4 stack at the
   endpoints so that the ALOC and ELOC elements are incorporated at the
   endpoint's stack; however, backwards compatibility must be preserved.
   Most applications will not be aware of the extensions while other IP-
   aware applications, such as Mobile IP, SIP, IPsec AH and so on (see
   Section 10.3) will suffer and cannot be used outside their ALOC realm
   when the hIPv4 framework is fully implemented, unless they are
   upgraded.  The reason is that the IP-aware applications depend upon
   the underlying network addressing structure, e.g., to identify an
   endpoint.

   Note that the applications used inside the local ALOC realm (e.g.,
   enterprise's private network) do not need to be upgraded -- neither
   in the intermediate nor in the long-term routing architecture.  The
   classical IPv4 framework is preserved in that only IP-aware
   applications used between ALOC realms need to be upgraded to support
   the hIPv4 header.

   Figure 1 shows a conceptual overview of the intermediate routing
   architecture.  When this architecture is in place, the ELOC space is
   no longer globally unique.  Instead, a regional allocation policy can
   be implemented.  For further details, see Appendix A.  The transition
   from the current routing architecture to the intermediate routing
   architecture is discussed in Appendix D.



























Frejborg                      Experimental                     [Page 14]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      Legend: *attachment point in the ALOC realm
              UER=Unique ELOC region
              EP=Endpoint

      |-------------------------------------------------------------|
      |            UER1          |       |           UER2           |
      |-------------------------------------------------------------|
      | Enterprise1 |    ISP1    |  ISP  |    ISP2    | Enterprise2 |
      |  ALOC Realm | ALOC Realm | Tier1 | ALOC Realm |  ALOC Realm |
      |             |            |       |            |             |
      |   *EP       |  *RBR      |       |  *RBR      |   *EP       |
      |    ELOC1    |   ALOC1    |       |   ALOC2    |    ELOC4    |
      |             |            |       |            |             |
      |             |   *EP      |       |   *EP      |             |
      |             |    ELOC2   |       |    ELOC3   |             |
      |             |            |       |            |             |
      |-------------|xxxxxxxxxxxxxx DFZ xxxxxxxxxxxxxx| ------------|
      |     RIB     |    RIB     |  RIB  |    RIB     |    RIB      |
      |             |            |       |            |             |
      |    ALOC1    |   ALOC1    | ALOC1 |   ALOC2    |    ALOC2    |
      |    ELOC1    |   ALOC2    | ALOC2 |   ALOC1    |    ELOC4    |
      |             |   ELOC2    |       |   ELOC3    |             |
      |             |   ELOC1    |       |   ELOC4    |             |
      |             |            |       |            |             |
      |-------------------------------------------------------------|

             Figure 1: Intermediate routing architecture of hIPv4

5.2.  Life of a hIPv4 Session



   This section provides an example of a hIPv4 session between two hIPv4
   endpoints: an initiator and a responder residing in different ALOC
   realms.

   When the hIPv4 stack is assembling the packet for transport, the
   hIPv4 stack shall decide if a classical IPv4 or a hIPv4 header is
   used based on the ALOC information received by a DNS reply.  If the
   initiator's local ALOC prefix equals the responder's ALOC prefix,
   there is no need to use the hIPv4 header for routing purposes,
   because both the initiator and responder reside in the local ALOC
   realm.  The packet is routed according to the prefixes in the IP
   header since the packet will not exit the local ALOC realm.  When the
   local ALOC prefix does not match the remote ALOC prefix, a hIPv4
   header must be assembled because the packet needs to be routed to a
   remote ALOC realm.






Frejborg                      Experimental                     [Page 15]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   A session between two endpoints inside an ALOC realm might use the
   locator header -- not for routing purposes, but to make use of
   Valiant Load-Balancing [VLB] for multipath-enabled transport
   protocols (see Section 11.1) or to make use of an identifier/locator
   split scheme (see Section 7).  When making use of VLB, the initiator
   adds the locator header to the packet and by setting the VLB-bits to
   01 or 11, indicating to the responder and intermediate routers that
   VLB is requested for the subflow.  Because this is an intra-ALOC
   realm session, there is no need to add ALOC and ELOC fields to the
   locator header, and thus the size of the locator header will be 4
   bytes.

   If an identifier/locator split scheme is applied for the session
   (intra-ALOC or inter-ALOC), the initiator must set the I-bit to 1 and
   make use of the Locator Header Length field.  Identifier/locator
   split scheme information is inserted into the locator header after
   the Locator Header Length field.

   How a hIPv4 session is established follows:

   1. The initiator queries the DNS server.  The hIPv4 stack notices
      that the local and remote ALOCs do not match and therefore must
      use the hIPv4 header for the session.  The hIPv4 stack of the
      initiator must assemble the packet by the following method:

      a. Set the local IP address from the API in the source address
         field of the IP header.

      b. Set the remote IP address from the API in the ELOC field of the
         locator header.

      c. Set the local ALOC prefix in the ALOC field of the locator
         header.

      d. Set the remote ALOC prefix in the destination address field of
         the IP header.

      e. Set the transport protocol value in the protocol field of the
         locator header and set the hIPv4 protocol value in the protocol
         field of the IP header.

      f. Set the desired parameters in the A-, I-, S-, VLB-, and L-
         fields of the locator header.

      g. Set the FI-bits of the locator header to 00.






Frejborg                      Experimental                     [Page 16]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      h. Calculate IP, locator, and transport protocol header checksums.
         The transport protocol header calculation does not include the
         locator header fields.  When completed, the packet is
         transmitted.

   2. The hIPv4 packet is routed throughout the Internet based on the
      value in the destination address field of the IP header.

   3. The hIPv4 packet will reach the closest RBR of the remote ALOC
      realm.  When the RBR notices that the value in the destination
      address of the IP header matches the local ALOC prefix, the RBR
      must:

      a. Verify that the received packet uses the hIPv4 protocol value
         in the protocol field of the IP header.

      b. Verify IP, locator, and transport protocol header checksums.
         The transport protocol header verification does not include the
         locator header fields.

      c. Replace the source address in the IP header with the ALOC
         prefix of the locator header.

      d. Replace the destination address in the IP header with the ELOC
         prefix of the locator header.

      e. Replace the ALOC prefix in the locator header with the
         destination address of the IP header.

      f. Replace the ELOC prefix in the locator header with the source
         address of the IP header.

      g. Set the S-field to 1.

      h. Decrease the Time to Live (TTL) value by one.

      i. Calculate IP, locator, and transport protocol header checksums.
         The transport header calculation does not include the locator
         header fields.

      j. Forward the packet according to the value in the destination
         address field of the IP header.

   4. The swapped hIPv4 packet is now routed inside the remote ALOC
      realm based on the new value in the destination address field of
      the IP header to the final destination.





Frejborg                      Experimental                     [Page 17]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   5. The responder receives the hIPv4 packet.

      a. The hIPv4 stack must verify that the received packet uses the
         hIPv4 protocol value in the protocol field of the IP header.

      b. Verify IP, locator, and transport protocol header checksums.
         The transport protocol header verification does not include the
         locator header fields.

   6. The hIPv4 stack of the responder must present the following to the
      extended IPv4 socket API:



      a. The source address of the IP header as the remote ALOC prefix.

      b. The destination address of the IP header as the local IP
         address.

      c. Verify that the received ALOC prefix of the locator header
         equals the local ALOC prefix.

      d. The ELOC prefix of the locator header as the remote IP address.

   The responder's application will respond to the initiator and the
   returning packet will take almost the same steps, which are steps 1
   to 6, as when the initiator started the session.  In step 1, the
   responder does not need to do a DNS lookup since all information is
   provided by the packet.

6.  Long-Term Routing Architecture

   The long-term routing architecture is established once the forwarding
   planes of private ALOC realms or service providers ALOC realms
   containing subscribers are upgraded.  The forwarding planes of
   transit DFZ routers do not need to be upgraded.  Why then would
   private network or service provider administrators upgrade their
   infrastructure?  There are two incentives:

   o  The overlay local ALOC exit routing topology (as discussed in
      Section 11) can be replaced by a peer-to-peer local ALOC exit
      routing topology, which is simpler to operate, thus decreasing
      operational expenditures.

   o  Locator freedom: Once the local ALOC realm is upgraded, the
      enterprise or service provider can use the full 32-bit ELOC
      address space to remove address space constraints and to design a
      well-aggregated routing topology with an overdimensioned ELOC
      allocation policy.




Frejborg                      Experimental                     [Page 18]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   When an enterprise or service provider upgrades the forwarding plane
   in their ALOC realm, the previous PI or PA address space allocation
   is released back to the RIR to be used for ALOC allocations in the
   GLB.

6.1.  Overview



   The swap service at the RBR was added to the framework in order to
   provide a smooth transition from the current IPv4 framework to the
   hIPv4 framework; a major upgrade of the current forwarding plane is
   avoided by the introduction of the swap service.  In the future, the
   swap service can be left "as is" in the ALOC realm, if preferred, or
   the swap service can be pushed towards the edge of the ALOC realm
   when routers are upgraded in their natural lifecycle process.

   Once an upgrade of a router is required because of, for example,
   increased demand for bandwidth, the modified forwarding plane might
   concurrently support IPv4 and hIPv4 forwarding -- and the swap
   service can be pushed towards the edge and in the future removed at
   the ALOC realm.  This is accomplished by adding an extension to the
   current routing protocols, both IGP and BGP.  When an RBR receives a
   hIPv4 packet where the value of the destination address field in the
   IP header matches the local ALOC prefix, the RBR will -- contrary to
   the tasks defined in Section 5.2, step 3 -- look up the ELOC field in
   the locator header and compare this prefix against the FIB.  If the
   next-hop entry is RBR-capable, the packet will be forwarded according
   to the ELOC prefix.  If the next-hop is a classical IPv4 router, the
   RBR must apply the tasks defined in Section 5.2, step 3 and, once
   completed, forward the packet according to the new value in the
   destination address field of the IP header.

   When all endpoints (that need to establish sessions outside the local
   ALOC realm) and infrastructure nodes in an ALOC realm are hIPv4-
   capable, there is no need to apply swap service for unicast sessions.
   Forwarding decisions can be based on information in the IP and
   locator headers.  In the local ALOC realm, packets are routed to
   their upstream anycast or unicast ALOC RBR according to the ALOC
   prefix in the locator header; local ALOC exit routing is applied
   against the local ALOC FIB.  Remote ELOC approach routing is applied
   against the ELOC FIB in the remote ALOC realm.

   Note that IP and transport protocol headers will remain intact
   (except for TTL values, since the RBR is a router); only FI and LH
   checksum values in the locator header will alternate in local ALOC
   exit routing mode and remote ELOC approach routing mode.






Frejborg                      Experimental                     [Page 19]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Figure 2 shows a conceptual overview of the long-term hIPv4 routing
   architecture.

   Legend: *attachment point in the ALOC realm
           UER=Unique ELOC region
           EP=Endpoint
           aRBR=anycast RBR
           uRBR=unicast RBR

     |-------------------------------------------------------------|
     |    UER1     |    UER2    |       |    UER3    |    UER4     |
     |-------------------------------------------------------------|
     | Enterprise1 |    ISP1    |  ISP  |    ISP2    | Enterprise2 |
     |  ALOC Realm | ALOC Realm | Tier1 | ALOC Realm |  ALOC Realm |
     |             |            |       |            |             |
     |   *EP       |  *aRBR     |       |  *aRBR     |   *EP       |
     |    ELOC1    |   ALOC1.1  |       |   ALOC2.1  |    ELOC4    |
     |             |            |       |            |             |
     |             *uRBR        |       |        uRBR*             |
     |             |ALOC1.2     |       |     ALOC2.2|             |
     |             |            |       |            |             |
     |             |   *EP      |       |   *EP      |             |
     |             |    ELOC2   |       |    ELOC3   |             |
     |             |            |       |            |             |
     |-------------|xxxxxxxxxxxxxx DFZ xxxxxxxxxxxxxx|-------------|
     |     RIB     |    RIB     |  RIB  |    RIB     |    RIB      |
     |             |            |       |            |             |
     |   ALOC1.2   |   ALOC1.1  | ALOC1 |   ALOC2.1  |   ALOC2.2   |
     |   ELOC1     |   ALOC1.2  | ALOC2 |   ALOC2.2  |   ELOC4     |
     |             |   ALOC2    |       |   ALOC1    |             |
     |             |   ELOC2    |       |   ELOC3    |             |
     |             |            |       |            |             |
     |-------------------------------------------------------------|

              Figure 2: Long-term routing architecture of hIPv4

   Also, the swap service for multicast can be removed when the
   forwarding planes are upgraded in all consequent ALOC realms.  The
   source's ALOC RBR sets the FI-bits to 11, and a Reverse Path
   Forwarding (RPF) check is hereafter applied against the ALOC prefix
   in the locator header.  Here, IP and transport protocol headers will
   not alternate.

   A long-term evolution will provide a 32x32 bit locator space.  The
   ALOC prefixes are allocated only to service providers; ELOC prefixes
   are only significant at a local ALOC realm.  An enterprise can use a
   32-bit locator space for its private network (the ALOC prefix is




Frejborg                      Experimental                     [Page 20]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   rented from the attached ISP), and an ISP can use a 32-bit ELOC space
   to provide Internet connectivity services for its directly attached
   customers (residential and enterprise).

6.2.  Exit, DFZ, and Approach Routing



   This section provides an example of a hIPv4 session between two hIPv4
   endpoints: an initiator in an ALOC realm where the forwarding plane
   has been upgraded to support the hIPv4 framework, and a responder
   residing in a remote ALOC realm with the classical IPv4 forwarding
   plane.

   When the forwarding plane at the local ALOC realm has been upgraded,
   the endpoints must be informed about it; that is, extensions to DHCP
   are needed or the endpoints are manually configured to be notified
   that the local ALOC realm is fully hIPv4 compliant.

   How a hIPV4 session is established follows:

   1. The initiator queries the DNS server.  The hIPv4 stack notices
      that the local and remote ALOCs do not match and therefore must
      use the hIPv4 header for the session.  The hIPv4 stack of the
      initiator must assemble the packet as described in Section 5.2,
      step 1, except for the following:

      g. Set the FI-bits of the locator header to 01.

   2. The hIPv4 packet is routed throughout the local ALOC realm
      according to the ALOC prefix of the locator header; local ALOC
      exit routing is applied.

   3. The hIPv4 packet will reach the closest RBR of the local ALOC
      realm.  When the RBR notices that the packet's ALOC prefix of the
      locator header matches the local ALOC prefix and the FI-bits are
      set to 01, the RBR must:

      a. Verify that the received packet uses the hIPv4 protocol value
         in the protocol field of the IP header.

      b. Verify the IP and locator header checksums.

      c. Set the FI-bits of the locator header to 00.

      d. Decrease the TTL value by one.

      e. Calculate IP and locator header checksums.





Frejborg                      Experimental                     [Page 21]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      f. Forward the packet according to the value in the destination
         address field of the IP header.

   4. The hIPv4 packet is routed to the responder as described in
      Section 5.2, steps 2 to 6.  DFZ routing is applied.

   5. The responder's application responds to the initiator and the
      returning packet takes almost the same steps as described in
      Section 5.2 except for:

   6. The hIPv4 packet will reach the closest RBR of the initiator's
      ALOC realm.  When the RBR notices that the value in the
      destination address field of the IP header matches the local ALOC
      prefix and the FI-bits are set to 00, the RBR must:

      a. Verify that the received packet uses the hIPv4 protocol value
         in the protocol field of the IP header.

      b. Verify the IP and locator header checksums.

      c. Set the FI-bits of the locator header to 10.

      d. Decrease the TTL value by one.

      e. Calculate IP and locator header checksums.

      f. Forward the packet according to the ELOC prefix of the locator
         header.

   7. The hIPv4 packet is routed throughout the initiator's ALOC realm
      according to the ELOC prefix of the locator header.  Remote ELOC
      approach routing is applied.

   8. The hIPv4 stack of the responder must present the following to the
      extended IPv4 socket API:



      a. The source address of the IP header as the remote IP address.

      b. The destination address of the IP header as the local ALOC
         prefix.

      c. The ALOC prefix of the locator header as the remote ALOC
         prefix.

      d. The ELOC prefix of the locator header as the local IP address.






Frejborg                      Experimental                     [Page 22]

RFC 6306               Hierarchical IPv4 Framework             July 2011


7.  Decoupling Location and Identification



   The design guidelines and rationale behind decoupling the location
   from identification are stated in [RFC6227].  Another important
   influence source is the report and presentations from the [Dagstuhl]
   workshop that declared "a future Internet architecture must hence
   decouple the functions of IP addresses as names, locators, and
   forwarding directives in order to facilitate the growth and new
   network-topological dynamisms of the Internet".

   Therefore, identifier elements need to be added to the hIPv4
   framework to provide a path for future applications to be able to
   remove the current dependency on the underlying network layer
   addressing scheme (local and remote IP address tuple).

   However, there are various ways to apply an identifier/locator split,
   as discussed in an [ID/loc_Split] presentation from the MobiArch
   workshop at Sigcomm 2008.  Thus, the hIPv4 framework will not propose
   or define a single identifier/locator split solution; a split can be
   achieved by, for example, a multipath transport protocol or by an
   identifier/locator database scheme such as HIP.  A placeholder has
   been added to the locator header so identifier/locator split schemes
   can be integrated into the hIPv4 framework.  But identifier/locator
   split schemes may cause privacy inconveniences, as discussed in
   [Mobility_&_Privacy].

   Multipath transport protocols, such as SCTP and the currently under
   development Multipath TCP (MPTCP) [RFC6182], are the most interesting
   candidates to enable an identifier/locator split for the hIPv4
   framework.  MPTCP is especially interesting from hIPv4's point of
   view; one of the main goals of MPTCP is to provide backwards
   compatibility with current implementations: hIPv4 shares the same
   goal.

   MPTCP itself does not provide an identifier/locator database scheme
   as HIP does.  Instead, MPTCP is proposing a token -- with local
   meaning -- to manage and bundle subflows under one session between
   two endpoints.  The token can be considered to have the
   characteristics of a session identifier, providing a generic cookie
   mechanism for the application layer and creating a session layer
   between the application and transport layers.  Thus, the use of a
   session identifier will provide a mechanism to improve mobility, both
   in site and endpoint mobility scenarios.

   Since the session identifier improves site and endpoint mobility,
   routing scalability is improved by introducing a hierarchical
   addressing scheme, why then add an identifier/locator database scheme
   to the hIPv4 framework?  Introducing an identifier/locator database



Frejborg                      Experimental                     [Page 23]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   scheme, as described in HIP, Identifier/Locator Network Protocol
   [ILNP] and Name-Based Sockets [NBS], might ease or remove the locator
   renumbering dependencies at firewalls that are used to scope security
   zones, but this approach would fundamentally change the currently
   deployed security architecture.

   However, combining an identifier/locator database scheme with DNS
   Security (DNSSEC) [RFC4033] is interesting.  Today, security zones
   are scoped by using locator prefixes in the security rule sets.
   Instead, a Fully Qualified Domain Name (FQDN) could be used in the
   rule sets and the renumbering of locator prefixes would no longer
   depend upon the security rule sets in firewalls.  Another interesting
   aspect is that an FQDN is and needs to be globally unique.  The ALOC
   prefix must be globally unique, but ELOC prefixes are only regionally
   unique and in the long-term only locally unique.  Nevertheless,
   combining identifier/locator database schemes with security
   architectures and DNSSEC needs further study.

   In order to provide multi-homing and mobility capabilities for single
   path transport protocols such as TCP and UDP, an identifier/locator
   database scheme is needed.  This scheme can also be used to create a
   bidirectional NAT traversal solution with a locator translation map
   consisting of private locator prefixes and public identifiers at the
   border router.

   The hIPv4 routing architecture provides only location information for
   the endpoints; that is, the ELOC describes how the endpoint is
   attached to the local network, and the ALOC prefixes describe how the
   endpoint is attached to the Internet.  Identifier/locator split
   schemes are decoupled from the routing architecture -- the
   application layer may or may not make use of an identifier/locator
   split scheme.

8.  ALOC Use Cases

   Several ALOC use cases are explored in this section.  As mentioned in
   Section 5.1, ALOC describes an area in the Internet that can span
   several autonomous systems (ASes), or if the area is equal to an AS
   you can say that the ALOC describes an AS.  When the ALOC describes
   an area, it is hereafter called an anycast ALOC.

   The ALOC can also be used to describe a specific node between two
   ALOC realms, e.g., a node installed between a private and an ISP ALOC
   realm, or between two private ALOC realms.  In this use case the ALOC
   describes an attachment point, e.g., where a private network is
   attached to the Internet.  This ALOC type is hereafter called a
   unicast ALOC.




Frejborg                      Experimental                     [Page 24]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   The main difference between anycast and unicast ALOC types is:

   o  In an anycast ALOC scenario, ELOC routing information is shared
      between the attached ALOC realms.

   o  In a unicast ALOC scenario, no ELOC routing information is shared
      between the attached ALOC realms.

   Unicast ALOC functionalities should not be deployed between private
   and ISP ALOC realms in the intermediate routing architecture -- it
   would require too many locators from the GLB space.  Instead, unicast
   ALOC functionality will be used to separate private ALOC realms.

   ALOC space is divided into two types, a globally unique ALOC space
   (a.k.a. GLB) that is installed in DFZ, and a private ALOC space that
   is used inside private networks.  Private ALOCs use the same locator
   space as defined in [RFC1918]; a private ALOC must be unique inside
   the private network and not overlap private ELOC prefixes.  Only ISPs
   should be allowed to apply for global ALOC prefixes.  For further
   discussion, see Appendix A.  The ISP should aggregate global ALOC
   prefixes as much as possible in order to reduce the size of the
   routing table in DFZ.

   When a user logs on to the enterprise's network, the endpoint will
   receive the following locator prefixes via provisioning means (e.g.,
   DHCP or manually configured):

   o  One ELOC prefix for each network interface.

   o  One private ALOC prefix due to

      -  The enterprise has recently been merged with another enterprise
         and overlapping ELOC spaces exist.

   o  Several private ALOC prefixes due to

      -  The enterprise network spans high-speed long-distance
         connections.  It is well-known that TCP cannot sustain high
         throughput for extended periods of time.  Higher throughput
         might be achieved by using multiple paths concurrently.

   o  One or several global ALOC prefixes.  These ALOCs describe how the
      enterprise network is attached to the Internet.

   As the user establishes a session to a remote endpoint, DNS is
   usually used to resolve remote locator prefixes.  DNS will return
   ELOC and ALOC prefixes of the remote endpoint.  If no ALOC prefixes
   are returned, a classical IPv4 session is initiated to the remote



Frejborg                      Experimental                     [Page 25]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   endpoint.  When ALOC prefixes are returned, the initiator compares
   the ALOC prefixes with its own local ALOC prefixes (that are provided
   via DHCP or manually configured).

   o  If the remote ALOC prefix is from the private ALOC space, the
      initiator will use the given private ALOC prefix for the session.

   Two use cases exist to design a network to use private ALOC
   functionality.  The remote endpoint is far away, leveraging high-
   speed long-distance connections, and in order to improve performance
   for the session a multipath transport protocol should be used.

   The other use case is when the remote endpoint resides in a network
   that recently has been merged and private ELOC [RFC1918] spaces
   overlap if no renumbering is applied.  One or several unicast ALOC
   solutions are needed in the network between the initiator and
   responder.  For long-distance sessions with no overlapping ELOC
   prefixes, anycast or unicast ALOC solutions can be deployed.

   A third use case follows; again the initiator compares returned ALOC
   prefixes from DNS with its own local ALOC prefixes:

   o  If the remote ALOC prefix is from the global ALOC space and the
      remote ALOC doesn't match the given global ALOC prefix, the
      initiator will use the given global ALOC prefix for the session.

   In this use case the remote endpoint resides outside the enterprise's
   private network, and the global remote ALOC prefixes indicate how the
   remote network is attached to the Internet.  When a multipath
   transport protocol is used, the subflows can be routed via separate
   border routers to the remote endpoint -- both at the local and remote
   sites, if both are multi-homed.  The initiator's egress packets in
   the local ALOC realm can be identified by the protocol value in the
   IP header, routed to an explicit path (e.g., MPLS LSP, L2TPv3 tunnel,
   etc.) based on the ALOC prefix in the locator header.  A local ALOC
   overlay exit routing scheme can be designed.  In the long-term
   routing architecture the overlay, the tunnel mechanism, can be
   removed; see Section 6.2.

   Figure 3 shows a conceptual diagram with two endpoints having a
   multipath session over a VPN connection and over the Internet (in the
   intermediate routing architecture).









Frejborg                      Experimental                     [Page 26]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Legend: *attachment point in the ALOC realm
           UER=Unique ELOC region
           EP=Endpoint
           aRBR=anycast RBR
           uRBR=unicast RBR
           BR=Border Router

     |-------------------------------------------------------------|
     |            UER1          |       |           UER2           |
     |-----------------------------------------------|-------------|
     | Enterprise1 |                                 | Enterprise2 |
     |  ALOC Realm |                                 |  ALOC Realm |
     |             |---------------------------------|             |
     |             |              VPN                |             |
     |             |           ALOC Realm            |             |
     |             *uRBR3                       uRBR4*             |
     |             |ALOC3                       ALOC4|             |
     |             |xxxxxxxxxxxX VPN RIB xxxxxxxxxxxx|             |
     |             |                                 |             |
     |             |           ALOC3 & ALOC4         |             |
     |             |---------------------------------|             |
     |   *EP1      |                                 |   *EP2      |
     |    ELOC1    |---------------------------------|    ELOC2    |
     |             |    ISP1    |  ISP  |    ISP2    |             |
     |             | ALOC Realm | Tier1 | ALOC Realm |             |
     |             |            |       |            |             |
     |          BR1*  *aRBR     |       |  *aRBR     *BR2          |
     |             |   ALOC1    |       |   ALOC2    |             |
     |             |            |       |            |             |
     |-------------|xxxxxxxxxxxxxx DFZ xxxxxxxxxxxxxx|-------------|
     |     RIB     |    RIB     |  RIB  |    RIB     |    RIB      |
     |             |            |       |            |             |
     |   ALOC1     |   ALOC1    | ALOC1 |   ALOC2    |   ALOC2     |
     |   ALOC3     |   ALOC2    | ALOC2 |   ALOC1    |   ALOC4     |
     |   ALOC4     |   ELOC1    |       |   ELOC2    |   ALOC3     |
     |   ELOC1     |            |       |            |   ELOC2     |
     |             |            |       |            |             |
     |-------------------------------------------------------------|

             Figure 3: Multi-pathing via VPN and the Internet

   The first subflow is established from the initiator (EP1) via uRBR3
   and uRBR4 (both use a private unicast ALOC prefix) to the responder
   (EP2).  Normal unicast forwarding is applied; ALOC prefixes of uRBR3
   and uRBR4 are installed in the routing tables of both the local and
   remote ALOC realms.  A second subflow is established via the
   Internet, that is, via BR1->BR2 to EP2. 0/0 exit routing is used to
   enter the Internet at both ALOC realms.



Frejborg                      Experimental                     [Page 27]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Note that ELOC prefixes can overlap since the local and remote ALOC
   realms reside in different ELOC regions and are separated by private
   unicast ALOC prefixes.

   The fourth use case is to leverage the private and global ALOC
   functionalities to be aligned with the design and implementation of
   [Split-DNS] solutions.

   The fifth use case is for residential users.  A residential user may
   use one or several ALOC prefixes, depending upon the service offer
   and network design of the ISP.  If the ISP prefers to offer advanced
   support for multipath transport protocols and local ALOC exit
   routing, the residential user is provided with several ALOC prefixes.
   The ALOC provided for residential users is taken from the GLB space
   and anycast ALOC functionality is applied.

9.  Mandatory Extensions



9.1.  Overview



   To implement the hierarchical IPv4 framework, some basic rules are
   needed:

   1. The DNS architecture must support a new extension; an A type
      Resource Record should be able to associate ALOC prefixes.

   2. An endpoint upgraded to support hIPv4 shall have information about
      the local ALOC prefixes; the local ALOC prefixes can be configured
      manually or provided via provisioning means such as DHCP.

   3. A globally unique IPv4 address block shall be reserved; this block
      is called the Global Locator Block (GLB).  A service provider can
      have one or several ALOC prefixes allocated from the GLB.

   4. ALOC prefixes are announced via current BGP to adjacent peers.
      They are installed in the RIB of the DFZ.  When the hIPV4
      framework is fully implemented, only ALOC prefixes are announced
      between the BGP peers in the DFZ.

   5. An ALOC realm must have one or several RBRs attached to it.  The
      ALOC prefix is configured as an anycast IP address on the RBR.
      The anycast IP address is installed to appropriate routing
      protocols in order to be distributed to the DFZ.

   6. The IPv4 socket API at endpoints must be extended to support local
      and remote ALOC prefixes.  The modified IPv4 socket API must be
      backwards compatible with the current IPv4 socket API.  The
      outgoing hIPv4 packet must be assembled by the hIPv4 stack with



Frejborg                      Experimental                     [Page 28]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      the local IP address from the socket as the source address and the
      remote ALOC prefix as the destination address in the IP header.
      The local ALOC prefix is inserted in the ALOC field of the locator
      header.  The remote IP address from the socket API is inserted in
      the ELOC field of the locator header.

9.2.  DNS Extensions



   Since the hierarchical IPv4 framework introduces an extended
   addressing scheme and because DNS serves as the "phone book" for the
   Internet, it is obvious that DNS needs a new Resource Record (RR)
   type to serve endpoints that are upgraded to support hIPv4.  Future
   RR types must follow the guidelines described in [RFC3597] and
   [RFC5395] with the following characteristics:

   o  Associated with the appropriate Fully Qualified Domain Name
      (FQDN), inserted in the NAME field.

   o  Assigned a new integer (QTYPE) in the TYPE field, to be assigned
      by IANA.

   o  The CLASS field is set to IN.

   o  The RDATA field is of an unknown type as defined in [RFC3597] and
      shall have the following format:

      o  Preference subfield: A 16-bit integer that specifies the
         preference given to this RR among others associated with a
         FQDN.  Lower values are preferred over higher values.

      o  ALOC subfield: A 32-bit integer that specifies the Area Locator
         of the associated FQDN.

                  +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
                  |                 Preference                    |
                  +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
                  |                                               |
                  |                    ALOC                       |
                  +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+

                         Figure 4: RDATA format of the ALOC RR

   Only endpoints that have been upgraded to support hIPv4 shall make
   use of the new ALOC RR. Also, there is no need to define a new ELOC
   RR because the A RR is used for that purpose when the ALOC RR is
   returned.





Frejborg                      Experimental                     [Page 29]

RFC 6306               Hierarchical IPv4 Framework             July 2011


9.3.  Extensions to the IPv4 Header



   Figure 5 shows how the locator header is added to the current IPv4
   header, creating a hIPv4 header.

   0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |Version|  IHL  |Type of Service|          Total Length         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Identification        |Flags|      Fragment Offset    |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Time to Live |    Protocol   |         Header Checksum       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                       Source Address                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Destination Address                       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Options                    |    Padding    |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |A|I|S| FI|VLB|L|    Protocol   |         LH Checksum           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Area Locator (optional)                    |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                 Endpoint Locator (optional)                   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |     LH Length (optional)    |        Padding (optional)       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                          Figure 5: hIPv4 header

   Version: 4 bits

      The Version field is identical to that of RFC 791.

   IHL: 4 bits

      The Internet Header Length field is identical to that of RFC 791.

   Type of Service: 8 bits

      The Type of Service is identical to that of RFC 791.

   Total Length:  16 bits

      The Total Length field is identical to that of RFC 791.





Frejborg                      Experimental                     [Page 30]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Identification:  16 bits

      The Identification field is identical to that of RFC 791.

   Flags:  3 bits

      The Flags field is identical to that of RFC 791.

   Fragment Offset:  13 bits

      The Fragment Offset field is identical to that of RFC 791.

   Time to Live:  8 bits

      The Time to Live field is identical to that of RFC 791.

   Protocol:  8 bits

      A new protocol number must be assigned for hIPv4.

   Header Checksum:  16 bits

      The Header Checksum field is identical to that of RFC 791.

   Source Address: 32 bits

      The Source Address field is identical to that of RFC 791.

   Destination Address: 32 bits

      The Destination Address field is identical to that of RFC 791.

   Options and Padding: Variable length

      The Options and Padding fields are identical to that of RFC 791.

   ALOC Realm Bit, A-bit: 1 bit

      When the initiator and responder reside in different ALOC realms,
      the A-bit is set to 1 and the Area and Endpoint Locator fields
      must be used in the locator header.  The size of the locator
      header is 12 bytes.  When the A-bit is set to 0, the initiator and
      responder reside within the same ALOC realm.  The Area and
      Endpoint Locator shall not be used in the locator header.  The
      size of the locator header is 4 bytes.






Frejborg                      Experimental                     [Page 31]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Identifier Bit, I-bit: 1 bit

      The identifier bit is set to 1 if the endpoint is using an
      identifier/locator split scheme within the locator header.  The
      identifier/locator split scheme must indicate by how much the size
      of the locator header is increased.  The Locator Header Length
      field is also added to the locator header.

   Swap Bit, S-bit: 1 bit

      The initiator sets the swap bit to 0 in the hIPv4 packet.  An RBR
      will set this bit to 1 when it is swapping the source and
      destination addresses of the IP header with the ALOC and ELOC
      prefixes of the locator header.

   Forwarding Indicator, FI-bits: 2 bits

      The purpose of the Forwarding Indicator (FI) field is to provide a
      mechanism for a future forwarding plane to identify which
      Forwarding Information Base (FIB) should be used for inter-ALOC
      realm sessions.  The new forwarding plane will remove the swap
      functionality of IP and locator header values for both unicast and
      multicast sessions.  The outcome is that the IP and transport
      protocol headers will remain intact and only FI and LH checksum
      values in the locator header will alternate.  The following values
      are defined:

         01: Local ALOC exit routing mode.  The initiator shall set the
         FI-bits to 01 and the ALOC prefix in the locator header is used
         to forward the packets to the RBR that is the owner of the
         local ALOC prefix.  The RBR shall change the FI-bits to 00.

         00: DFZ routing mode.  The local ALOC RBR shall forward the
         packets according to the value in the destination address field
         of the IP header.  The DFZ routers shall forward the packets
         based on the value in the destination address field of the IP
         header unless the destination address matches the local ALOC
         prefix.  When this situation occurs, the packet enters the
         remote ALOC realm and the remote RBR shall change the FI-bits
         to 10.

         10: Remote ELOC approach routing mode.  The remote ALOC RBR and
         following routers shall forward the packets based on the ELOC
         prefix in the locator header.







Frejborg                      Experimental                     [Page 32]

RFC 6306               Hierarchical IPv4 Framework             July 2011


         11: Inter-ALOC RPF check mode.  The local ALOC RBR changes the
         FI-bits to 11 and the following inter-ALOC routers on the
         shared tree shall apply the RPF check against the ALOC prefix
         in the locator header.

   Valiant Load-Balancing, VLB-bits: 2 bits (optional, subject for
   further research)

      The purpose of the Valiant Load-Balancing field is to provide a
      mechanism for multipath-enabled transport protocols to request
      explicit paths in the network for subflows, which are component
      parts of a session between two endpoints.  The subflow path
      request can be set as follows:

         00: Latency-sensitive application.  Only one single subflow
         (multipath not applied), the shortest path through the network
         is requested.



         01: First subflow.  The shortest path or Valiant Load-Balancing
         might be applied.



         11: Next subflow(s).  Valiant Load-Balancing should be applied



   Load-Balanced, L-bit: 1 bit (optional, subject for further research)

      The initiator must set the L-bit to zero.  A Valiant Load-
      Balancing-capable node can apply VLB switching for the session if
      the value is set to zero; if the value is set to 1, VLB switching
      is not allowed.  When VLB switching is applied for the session,
      the node applying the VLB algorithm must set the value to 1.

   Protocol: 8 bits

      The Protocol field is identical to that of RFC 791.

   Locator Header Checksum: 16 bits

      A checksum is calculated for the locator header only.  The
      checksum is computed at the initiator, recomputed at the RBR, and
      verified at the responder.  The checksum algorithm is identical to
      that of RFC 791.

   Area Locator (optional): 32 bits

      The Area Locator is an IPv4 address assigned to locate an ALOC
      realm in the Internet.  The ALOC is assigned by an RIR to a
      service provider.  The ALOC is globally unique because it is
      allocated from the GLB.



Frejborg                      Experimental                     [Page 33]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Endpoint Locator (optional): 32 bits

      The Endpoint Locator is an IPv4 address assigned to locate an
      endpoint in a local network.  The ELOC block is assigned by an RIR
      to a service provider or to an enterprise.  In the intermediate
      routing architecture the ELOC block is only unique in a
      geographical region.  The final policy of uniqueness shall be
      defined by the RIRs.  In the long-term routing architecture the
      ELOC block is no longer assigned by an RIR; it is only unique in
      the local ALOC realm.

   Locator Header Length (optional): 16 bits

      The Locator Header Length is the total length of the locator
      header.  Locator Header Length is applied when the identifier bit
      is set to 1.  Identifier/locator split scheme parameters are
      inserted into the locator header after this field.

   Padding (optional): variable

      The locator header padding is used to ensure that the locator
      header ends on a 32-bit boundary.  The padding is zero.

10.  Consequences



10.1.  Overlapping Local and Remote ELOC Prefixes/Ports



   Because an ELOC prefix is only significant within the local ALOC
   realm, there is a slight possibility that a session between two
   endpoints residing in separate ALOC realms might use the same local
   and remote ELOC prefixes.  But the session is still unique because
   the two processes communicating over the transport protocol form a
   logical session that is uniquely identifiable by the 5-tuple
   involved, by the combination of <protocol, local IP address, local
   port, remote IP address, remote port>.

   The session might no longer be unique when two initiators with the
   same local ELOC prefix residing in two separate ALOC realms are
   accessing a responder located in a third ALOC realm.  In this
   scenario, the possibility exists that the initiators will use the
   same local port value.  This situation will cause an "identical
   session situation" for the application layer.

   To overcome this scenario, the hIPv4 stack must accept only one
   unique session with the help of the ALOC information.  If there is an
   "identical session situation", i.e., both initiators use the same
   values in the 5-tuple <protocol, local IP address, local port, remote
   IP address, remote port>, the hIPv4 stack shall allow only the first



Frejborg                      Experimental                     [Page 34]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   established session to continue.  The following sessions must be
   prohibited and the initiator is informed by ICMP notification about
   the "identical session situation".

   MPTCP introduces a token that is locally significant and currently
   defined as 32 bits long.  The token will provide a sixth tuple for
   future applications to identify and verify the uniqueness of a
   session.  Thus, the probability to have an "identical session
   situation" is further reduced.  By adding an identifier/locator
   database scheme to the hIPv4 framework, the "identical session
   situation" is completely removed.

10.2.  Large Encapsulated Packets



   Adding the locator header to an IPv4 packet in order to create a
   hIPv4 packet will increase the size of it, but since the packet is
   assembled at the endpoint it will not add complications of the
   current Path MTU Discovery (PMTUD) mechanism in the network.  The
   intermediate network between two endpoints will not see any
   difference in the size of packets; IPv4 and hIPv4 packet sizes are
   the same from the network point of view.

10.3.  Affected Applications



   There are several applications that insert IP address information to
   the payload of a packet.  Some applications use the IP address
   information to create new sessions or for identification purposes.
   Some applications collect IP address information to be used as
   referrals.  This section tries to list the applications that need to
   be enhanced; however, this is by no means a comprehensive list.  The
   applications can be divided into five main categories:

   o  Applications based on raw sockets - a raw socket receives packets
      containing the complete header, in contrast to the other sockets
      that only receive the payload.

   o  Applications needed to enable the hIPv4 framework, such as DNS and
      DHCP databases, which must be extended to support ALOC prefixes.

   o  Applications that insert IP addresses into the payload or use the
      IP address for setting up new sessions or for some kind of
      identification or as referrals.  An application belonging to this
      category cannot set up sessions to other ALOC realms until
      extensions have been incorporated.  Within the local ALOC realm
      there are no restrictions since the current IPv4 scheme is still
      valid.  The following applications have been identified:





Frejborg                      Experimental                     [Page 35]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      -  SIP: IP addresses are inserted in the SDP offers/answers, XML
         body, Contact, Via, maddr, Route, Record-Route SIP headers.

      -  Mobile IP: the mobile node uses several IP addresses during the
         registration process.

      -  IPsec AH: designed to detect alterations at the IP packet
         header.

      -  RSVP: Resource Reservation Protocol (RSVP) messages are sent
         hop-by-hop between RSVP-capable routers to construct an
         explicit path.

      -  ICMP: notifications need to be able to incorporate ALOC
         information and assemble the hIPv4 header in order to be routed
         back to the source.

      -  Source Specific Multicast: the receiver must specify the source
         address.

      -  IGMPv3: a source-list is included in the IGMP reports.

   o  Applications related to security, such as firewalls, must be
      enhanced to support ALOC prefixes.

   o  Applications that will function with FQDN, but many use IP
      addresses instead, such as ping, traceroute, telnet, and so on.
      The CLI syntax needs to be upgraded to support ALOC and ELOC
      information via the extended socket API.

   At first glance, it seems that a lot of applications need to be re-
   engineered and ported, but the situation is not all that bad.  The
   applications used inside the local ALOC realm (e.g., an enterprise's
   private network) do not need to be upgraded, neither in the
   intermediate nor in the long-term architecture.  The classical IPv4
   framework is preserved.  Only IP-aware applications used between ALOC
   realms need to be upgraded to support the hIPv4 header.  IPv6 has the
   definitions in place of the applications mentioned above, but the
   migration of applications from IPv4 to IPv6 can impose some capital
   expenditures for enterprises, especially if the applications are
   customized or homegrown; see [Porting_IPv4].

   As stated earlier, hIPv4 does not require to port applications used
   inside a private network.  The conclusion is that, whatever next
   generation architecture is deployed, some applications will suffer,
   either during the transition period or when being re-engineered in
   order to be compatible with the new architecture.




Frejborg                      Experimental                     [Page 36]

RFC 6306               Hierarchical IPv4 Framework             July 2011


10.4.  ICMP



   As long as the ICMP request is executed inside the local ALOC realm,
   the normal IPv4 ICMP mechanism can be used.  As soon as the ICMP
   request exits the local ALOC realm, the locator header shall be used
   in the notifications.  Therefore, extensions to the ICMP shall be
   implemented.  These shall be compatible with [RFC4884] and support
   ALOC and ELOC information.

10.5.  Multicast



   Since local ELOC prefixes are only installed in the routing table of
   the local ALOC realm, there is a constraint with Reverse Path
   Forwarding (RPF) that is used to ensure loop-free forwarding of
   multicast packets.  The source address of a multicast group (S,G) is
   used against the RPF check.  The address of the source can no longer
   be used as a RPF checkpoint outside the local ALOC realm.

   To enable RPF globally for an (S,G), the multicast-enabled RBR (mRBR)
   must at the source's ALOC realm replace the value of the source
   address field in the IP header with the local ALOC prefix for inter-
   ALOC multicast streams.  This can be achieved if the local RBR acts
   also as an anycast Rendezvous Point with MSDP and PIM capabilities.
   With these functionalities the RBR becomes a multicast-enabled RBR
   (mRBR).  The source registers at the mRBR and a source tree is
   established between the source and the mRBR.  When an inter-ALOC
   realm receiver subscribes to the multicast group, the mRBR has to
   swap the hIPv4 header in the following way:

   a. Verify that the received packet uses the hIPv4 protocol value in
      the protocol field of the IP header.

   b. Verify IP, locator, and transport protocol header checksums.

   c. Replace the source address in the IP header with the local ALOC
      prefix.

   d. Set the S-field to 1.

   e. Decrease the TTL value by one.

   f. Calculate IP, locator, and transport protocol header checksums.
      Transport protocol header calculations do not include the locator
      header fields.

   g. Forward the packet to the shared multicast tree.





Frejborg                      Experimental                     [Page 37]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   In order for the mRBR to function as described above, the source must
   assemble the multicast hIPv4 packet in the following way:

   a. Set the local IP address (S) from the API in the source address
      field of the IP header and in the ELOC field of the locator
      header.

   b. Set the multicast address (G) from the API in the destination
      address field of the IP header.

   c. Set the local ALOC prefix in the ALOC field of the locator header.

   d. Set the transport protocol value in the protocol field of the
      locator header and the hIPv4 protocol value in the protocol field
      of the IP header.

   e. Set the desired parameters in the A-, I-, S-, VLB-, and L-fields
      of the locator header.

   f. Set the FI-bits of the locator header to 00.

   g. Calculate IP, locator, and transport protocol header checksums.
      Transport protocol header calculations do not include the locator
      header fields.  When completed, the packet is transmitted.

   The downstream routers from the mRBR towards the receiver will use
   the source address (which is the source's ALOC prefix after the mRBR)
   in the IP header for RPF verification.  In order for the receiver to
   create Real-time Transport Control Protocol (RTCP) receiver reports,
   all information is provided in the hIPv4 header of the packet.

   Because Source Specific Multicast (SSM) and IGMPv3 use IP addresses
   in the payload, both protocols need to be modified to support the
   hIPv4 framework.

11.  Traffic Engineering Considerations

   When the intermediate phase of the hIPv4 framework is fully
   implemented, ingress load balancing to an ALOC realm can be
   influenced by the placement of RBRs at the realm; an RBR provides a
   shortest path scheme.  Also, if RIR policies allow, a service
   provider can have several ALOCs assigned.  Hence, traffic engineering
   and filtering can be done with the help of ALOC prefixes.  For
   example, sensitive traffic can be aggregated under one ALOC prefix
   that is not fully distributed into the DFZ.






Frejborg                      Experimental                     [Page 38]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   If needed, an ALOC traffic engineering solution between ALOC realms
   might be developed, to create explicit paths that can be engineered
   via specific ALOC prefixes.  For example, develop a mechanism similar
   to the one described in [Pathlet_Routing].  Further studies are
   needed; first it should be evaluated whether there is demand for such
   a solution.

   Ingress load balancing to a private remote ALOC realm (remote site)
   is influenced by how many attachment points to the Internet the site
   uses and where the attachment points are placed at the site.  In
   order to apply local ALOC exit routing, e.g., from a multi-homed
   site, some new network nodes are needed between the initiator and the
   border routers of the site.

   In the intermediate routing architecture this is achieved by using
   overlay architectures such as MPLS LSP, L2TPv3 tunnels, etc.  The new
   network node(s) shall be able to identify hIPv4 packets, based on the
   protocol field in the IP header, and switch the packets to explicit
   paths based on the ALOC prefix in the locator header.  In the long-
   term routing architecture the overlay solution is replaced with a new
   forwarding plane; see Section 6.2.

   Together with a multipath transport protocol, the subflows can be
   routed via specific attachment points, that is, border routers
   sitting between the private local/remote ALOC realms (multi-homed
   sites) and the Internet.  Multi-homing becomes multi-pathing.  For
   details, see Appendix B.

11.1.  Valiant Load-Balancing



   The use of multipath-enabled transport protocols opens up the
   possibility to develop a new design methodology of backbone networks,
   based on Valiant Load-Balancing [VLB].  If two sites that are
   connected with a single uplink to the Internet, and the endpoints are
   using multipath-enabled transport protocols and are attached to the
   network with only one interface/ELOC-prefix, both subflows will most
   likely take the shortest path throughout the Internet.  That is, both
   subflows are established over the same links and when there is
   congestion on a link or a failure of a link, both subflows might
   simultaneously drop packets.  Thus, the benefit of multi-pathing is
   lost.

   The "subflows-over-same-links" scenario can be avoided if the
   subflows are traffic engineered to traverse the Internet on different
   paths, but this is difficult to achieve by using classical traffic
   engineering, such as IGP tuning or MPLS-based traffic engineering.
   By adding a mechanism to the locator header, the "subflows-over-same-
   links" scenario might be avoided.



Frejborg                      Experimental                     [Page 39]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   If the RBR functionality is deployed on a Valiant Load-Balancing
   enabled backbone node -- hereafter called vRBR -- and the backbone
   nodes are interconnected via logical full meshed connections, Valiant
   Load-Balancing can be applied for the subflows.  When a subflow has
   the appropriate bits set in the VLB-field of the locator header, the
   first ingress vRBR shall do VLB switching of the subflow.  That is,
   the ingress vRBR is allowed to do VLB switching of the subflow's
   packets if the VLB-bits are set to 01 or 11, the L-bit is set to 0,
   and the local ALOC prefix of the vRBR matches the ALOC-field's
   prefix.  If there are no ALOC and ELOC fields in the locator header,
   but the other fields' values are set as described above, the vRBR
   should apply VLB switching as well for the subflow -- because it is
   an intra-ALOC realm subflow belonging to a multipath-enabled session.

   With this combination of parameters in the locator header, the
   subflow is VLB switched only at the first ALOC realm and the subflows
   might be routed throughout the Internet on different paths.  If VLB
   switching is applied at every ALOC realm, this would most likely add
   too much latency for the subflows.  The VLB switching at the first
   ALOC realm will not separate the subflows on the first and last mile
   links (site with a single uplink).  If the subflows on the first and
   last mile link need to be routed on separate links, the endpoints
   should be deployed in a multi-homed environment.  Studies on how
   Valiant Load-Balancing is influencing traffic patterns between
   interconnected VLB [iVLB] backbone networks have been done.
   Nevertheless, more studies are needed regarding Valiant Load-
   Balancing scenarios.

12.  Mobility Considerations



   This section considers two types of mobility solutions: site mobility
   and endpoint mobility.

   Site mobility:

   Today, classical multi-homing is the most common solution for
   enterprises that wish to achieve site mobility.  Multi-homing is one
   of the key findings behind the growth of the DFZ RIB; see [RFC4984],
   Sections 2.1 and 3.1.2.  The hIPv4 framework can provide a solution
   for enterprises to have site mobility without the requirement of
   implementing a classical multi-homed solution.

   One of the reasons to deploy multi-homing is to avoid renumbering of
   the local infrastructure when an upstream ISP is replaced.  Thus,
   today, PI-address blocks are deployed at enterprises.  In the
   intermediate routing architecture, an enterprise is allocated a
   regional PI ELOC block (for details, see Appendix A) that is used for
   internal routing.  The upstream ISP provides an ALOC prefix that



Frejborg                      Experimental                     [Page 40]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   describes how the enterprise's network is connected to the Internet.
   If the enterprise wishes to switch to another ISP, it only changes
   the ALOC prefix at endpoints, from the previous ISP's ALOC prefix to
   the new ISP's ALOC prefix, without connectivity interruptions in the
   local network since the ALOC prefix is only used for Internet
   connectivity -- several ALOCs can be used simultaneously at the
   endpoints; thus, a smooth migration from one ISP to another is
   possible.  In the long-term routing architecture, when the forwarding
   plane is upgraded, the regional PI ELOC block is returned to the RIR
   and the enterprise can use a full 32-bit ELOC space to design the
   internal routing topology.

   An enterprise can easily become multi-homed or switch ISPs.  The
   local ELOC block is used for internal routing and upstream ISPs
   provide their ALOC prefixes for Internet connectivity.  Multi-homing
   is discussed in detail in Appendix B.

   Endpoint mobility:

   As said earlier, MPTCP is the most interesting identifier/locator
   split scheme to solve endpoint mobility scenarios.  MPTCP introduces
   a token, which is locally significant and currently defined as 32
   bits long.  The token will provide a sixth tuple to identify and
   verify the uniqueness of a session.  This sixth tuple -- the token --
   does not depend upon the underlying layer, the IP layer.  The session
   is identified with the help of the token and thus the application is
   not aware when the locator parameters are changed, e.g., during a
   roaming situation, but it is required that the application is not
   making use of ELOC/ALOC information.  In multi-homed scenarios, the
   application can make use of ELOC information, which will not change
   if the endpoint is fixed to the location.

   Security issues arise: the token can be captured during the session
   by, for example, a man-in-the-middle attack.  These attacks can be
   mitigated by applying [tcpcrypt], for example.  If the application
   requires full protection against man-in-the-middle attacks, the user
   should apply the Transport Layer Security Protocol (TLS) [RFC5246]
   for the session.

   The most common endpoint mobility use case today is that the
   responder resides in the fixed network and the initiator is mobile.
   Thus, MPTCP will provide roaming capabilities for the mobile
   endpoint, if both endpoints are making use of the MPTCP extension.
   However, in some use cases, the fixed endpoint needs to initialize a
   session to a mobile responder.  Therefore, Mobile IP (MIP) [RFC5944]
   should incorporate the hIPv4 extension -- MIP provides a rendezvous
   service for the mobile endpoints.




Frejborg                      Experimental                     [Page 41]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Also, many applications provide rendezvous services for their users,
   e.g., SIP, peer-to-peer, Instant Messaging services.  A generic
   rendezvous service solution can be provided by an identifier/locator
   database scheme, e.g., HIP, ILNP, or NBS.  If desired, the user
   (actually the application) can make use of one of these rendezvous
   service schemes, such as extended MIP, some application-specific
   rendezvous services, or a generic rendezvous service -- or some
   combination of them.

   The hIPv4 framework will not define which identifier/locator split
   solution should be used for endpoint mobility.  The hIPv4 framework
   is focusing on routing scalability and supports several
   identifier/locator split solutions that can be exploited to develop
   new services, with the focus on endpoint mobility.

13.  Transition Considerations



   The hIPv4 framework is not introducing any new protocols that would
   be mandatory for the transition from IPv4 to hIPv4; instead,
   extensions are added to existing protocols.  The hIPv4 framework
   requires extensions to the current IPv4 stack, to infrastructure
   systems, and to some applications that use IP address information,
   but the current forwarding plane in the Internet remains intact,
   except that a new forwarding element (the RBR) is required to create
   an ALOC realm.

   Extensions to the IPv4 stack, to infrastructure systems, and to
   applications that make use of IP address information can be deployed
   in parallel with the current IPv4 framework.  Genuine hIPv4 sessions
   can be established between endpoints even though the current
   unidimensional addressing structure is still present.

   When will the unidimensional addressing structure be replaced by a
   hierarchical addressing scheme and a fourth hierarchy added to the
   routing architecture?  The author thinks there are two possible
   tipping points:

   o  When the RIB of DFZ is getting close to the capabilities of
      current forwarding planes.  Who will pay for the upgrade?  Or will
      the service provider only accept ALOC prefixes from other service
      providers and avoid capital expenditures?

   o  When the depletion of IPv4 addresses is causing enough problems
      for service providers and enterprises.

   The biggest risk and reason why the hIPv4 framework will not succeed
   is the very short time frame until the expected depletion of the IPv4
   address space occurs -- actually the first RIR has run out of IPv4



Frejborg                      Experimental                     [Page 42]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   addresses during the IESG review process of this document (April
   2011).  Also, will enterprises give up their global allocation of the
   current IPv4 address block they have gained, as an IPv4 address block
   has become an asset with an economical value.

   The transition requires the upgrade of endpoint's stack, and this is
   a drawback compared to the [CES] architectures proposed in [RFC6115].
   A transition to an architecture that requires the upgrade of
   endpoint's stack is considerably slower than an architecture that
   requires only upgrade of some network nodes.  But the transition
   might not be as slow or challenging at it first seems since hIPv4 is
   an evolution of the current deployed Internet.

   o  Not all endpoints need to be upgraded; the endpoints that do not
      establish sessions to other ALOC realms can continue to make use
      of the classical IPv4 framework.  Also, legacy applications that
      are used only inside a local ALOC realm do not need to be ported
      to another framework.  For further details, see Appendix C.

   o  Upgrading endpoint's stack, e.g., at critical or complicated
      systems, will definitely take time; thus, it would be more
      convenient to install a middlebox in front of such systems.  It is
      obvious that the hIPv4 framework needs a middlebox solution to
      speed up the transition; combining CES architectures with the
      hIPv4 framework might produce such a middlebox.  For further
      details, see Appendix D.

   o  The framework is incrementally deployable.  Not all endpoints in
      the Internet need to be upgraded before the first IPv4 block can
      be released from a globally unique allocation status to a
      regionally unique allocation status.  That is, to achieve ELOC
      status for the prefixes used in a local network in the
      intermediate routing architecture, see Appendix D.  An ALOC realm
      that wishes to achieve local unique status for its ELOC block in
      the long-term routing architecture does not need to wait for other
      ALOC realms to proceed to the same level simultaneously.  It is
      sufficient that the other ALOC realms have achieved the
      intermediate routing architecture status.  For further details,
      see Section 6.

14.  Security Considerations



   Because the hIPv4 framework does not introduce other network
   mechanisms than a new type of border router to the currently deployed
   routing architecture, the best current practices for securing ISP
   networks are still valid.  Since the DFZ will no longer contain ELOC
   prefixes, there are some benefits and complications regarding
   security that need to be taken into account.



Frejborg                      Experimental                     [Page 43]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   The hijacking of a single ELOC prefix by longest match from another
   ALOC realm is no longer possible because the prefixes are separated
   by a locator, the ALOC.  To carry out a hijack of a certain ELOC
   prefix, the whole ALOC realm must be routed via a bogus ALOC realm.
   Studies should be done with the Secure Inter-Domain Routing (SIDR)
   working group to determine whether the ALOC prefixes can be protected
   from hijacking.

   By not being able to hijack a certain ELOC prefix, there are some
   implications when mitigating distributed denial-of-service (DDoS)
   attacks.  This implication occurs especially in the long-term routing
   architecture, e.g., when a multi-homed enterprise is connected with
   unicast ALOC RBRs to the ISPs.

   One method used today to mitigate DDoS attacks is to inject a more
   specific prefix (typically host prefix) to the routing table so that
   the victim of the attack is "relocated", i.e., a sinkhole is created
   in front of the victim.  The sinkhole may separate bogus traffic from
   valid traffic or analyze the attack.  The challenge in the long-term
   routing architecture is how to reroute a specific ELOC prefix of the
   multi-homed enterprise when the ELOC prefix cannot be installed in
   the ISP's routing table.

   Creating a sinkhole for all traffic designated to an ALOC realm might
   be challenging and expensive, depending on the size of the multi-
   homed enterprise.  To have the sinkhole at the enterprise's ALOC
   realm may saturate the connections between the enterprise and ISPs,
   thus this approach is not a real option.

   By borrowing ideas from a service-centric networking architecture
   [SCAFFOLD], a sinkhole service can be created.  An example of how a
   distributed sinkhole service can be designed follows:

      a. A firewall (or similar node) at the victim's ALOC realm
         discovers an attack.  The security staff at the enterprise
         realizes that the amount of the incoming traffic caused by the
         attack is soon saturating the connections or other resources.
         Thus, the staff informs the upstream ISPs of the attack, also
         about the victim's ALOC prefix X and ELOC prefix Y.

      b. The ISP reserves the resources for the sinkhole service.  These
         resources make use of ALOC prefix Z; the resources are
         programmed with a service ID and the victim's X and Y prefixes.
         The ISP informs the victim's security staff of the service ID.
         The ISP applies a NAT rule on their RBRs and/or hIPv4-enabled
         routers.  The NAT rule replaces the destination address in the
         IP header of packets with Z when the destination address of the
         IP header matches X and the ELOC prefix of the locator header



Frejborg                      Experimental                     [Page 44]

RFC 6306               Hierarchical IPv4 Framework             July 2011


         matches Y.  Also, the service ID is inserted to the locator
         header; the service ID acts as a referral for the sinkhole.  It
         is possible that the sinkhole serves several victims; thus, a
         referral is needed.  PMTUD issues must be taken into account.

      c. The victim's inbound traffic is now routed at the RBRs and/or
         hIPv4-enabled routers to the sinkhole(s); the traffic is
         identified by the service ID.  Bogus traffic is discarded at
         the sinkhole, for valid traffic the value of the destination
         address in the IP header Z is replaced with X.  By using a
         service ID in the analyzed packets, the enterprise is informed
         that the packets containing service ID are valid traffic and
         allowed to be forwarded to the victim.  It might be possible
         that not all upstream ISPs are redirecting traffic to the
         distributed sinkholes.  Thus, traffic that does not contain the
         agreed service ID might be bogus.  Also, by inserting a service
         ID to the valid packets, overlay solutions between the routers,
         sinkholes, and victim can be avoided.  In case the valid packet
         with a service ID traverses another RBR or hIPv4-enabled router
         containing the same NAT rule, that packet is not rerouted to
         the sinkhole.  The enterprise shall ensure that the victim does
         not use the service ID in its replies -- if the attacker
         becomes aware of the service ID, the sinkhole is disarmed.

   Today, traffic is sent to sinkholes by injecting host routes into the
   routing table.  This method can still be used inside an ALOC realm
   for intra-ALOC attacks.  For attacks spanning over several ALOC
   realms new methods are needed; one example is described above.  It is
   desirable that the RBR and hIPv4-enabled routers are capable of
   applying NAT rules and inserting service ID to selected packets in
   the forwarding plane.

15.  Conclusions



   This document offers a high-level overview of the hierarchical IPv4
   framework that could be built in parallel with the current Internet
   by implementing extensions at several architectures.  Implementation
   of the hIPv4 framework will not require a major service window break
   in the Internet or at the private networks of enterprises.
   Basically, the hIPv4 framework is an evolution of the current IPv4
   framework.

   The transition to hIPv4 might be attractive for enterprises since the
   hIPv4 framework does not create a catch-22 situation, e.g., when
   should an application used only inside the private network be ported
   from IPv4 to IPv6?  Also, what is the business justification for
   porting the application to IPv6?  Another matter is that when an




Frejborg                      Experimental                     [Page 45]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   IPv4/v6 dual-stack solution is used it could impose operational
   expenditures, especially with rule sets at firewalls -- both in front
   of servers and at clients.

   If an enterprise chooses to deploy hIPv4, however, the legacy
   applications do not need to be ported because hIPv4 is backwards
   compatible with the classical IPv4 framework.  This means lower costs
   for the enterprise, and an additional bonus is the new stack's
   capabilities to better serve mobility use cases.

   But the enterprise must take the decision soon and act promptly,
   because the IPv4 address depletion is a reality in the very near
   future.  If the decision is delayed, IPv6 will arrive, and then,
   sooner or later, the legacy applications will need to be ported.

   However, though this document has focused only on IPv4, a similar
   scheme can be deployed for IPv6 in the future, that is, creating a
   64x64 bit locator space.  But some benefits would have been lost at
   the time this document was written, such as:

      o  Backwards compatibility with the current Internet and therefore
         no smooth migration plan is gained.

      o  The locator header, including ALOC and ELOC prefixes, would
         have been larger, 160 bits versus 96 bits.  And the identifier
         (EUI-64) would always have been present, which can be
         considered as pros or cons, depending upon one's view of the
         privacy issue, as discussed in [RFC4941] and in
         [Mobility_& _Privacy].

   If an enterprise prefers hIPv4 (e.g., due to gaining additional IPv4
   addresses and smooth migration capabilities), there is an
   unintentional side effect (from the enterprise's point of view) on
   the routing architecture of the Internet; multi-homing becomes multi-
   pathing, and an opportunity opens up for the service providers to
   create an Internet routing architecture that holds less prefixes and
   generates less BGP updates in DFZ than the current Internet.

   The hIPv4 framework is providing a new hierarchy in the routing
   subsystem and is complementary work to multipath-enabled transport
   protocols (such as MPTCP and SCTP) and service-centric networking
   architectures (such as SCAFFOLD).  End users and enterprises are not
   interested in routing issues in the Internet; instead, a holistic
   view should be applied on the three disciplines with a focus on new
   service opportunities and communicated to the end users and
   enterprises.  Then perhaps the transition request to a new routing
   architecture will be accepted and carried out.  However, more work is
   needed to accomplish a holistic framework of the three disciplines.



Frejborg                      Experimental                     [Page 46]

RFC 6306               Hierarchical IPv4 Framework             July 2011


16.  References



16.1.  Normative References



   [RFC1385]   Wang, Z., "EIP: The Extended Internet Protocol", RFC
               1385, November 1992.

   [RFC1812]   Baker, F., Ed., "Requirements for IP Version 4 Routers",
               RFC 1812, June 1995.

   [RFC1918]   Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G.,
               and E. Lear, "Address Allocation for Private Internets",
               BCP 5, RFC 1918, February 1996.

   [RFC2119]   Bradner, S., "Key words for use in RFCs to Indicate
               Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC3031]   Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol
               Label Switching Architecture", RFC 3031, January 2001.

   [RFC4033]   Arends, R., Austein, R., Larson, M., Massey, D., and S.
               Rose, "DNS Security Introduction and Requirements", RFC
               4033, March 2005.

   [RFC4601]   Fenner, B., Handley, M., Holbrook, H., and I. Kouvelas,
               "Protocol Independent Multicast - Sparse Mode (PIM-SM):
               Protocol Specification (Revised)", RFC 4601, August 2006.

   [RFC4884]   Bonica, R., Gan, D., Tappan, D., and C. Pignataro,
               "Extended ICMP to Support Multi-Part Messages", RFC 4884,
               April 2007.

   [RFC5246]   Dierks, T. and E. Rescorla, "The Transport Layer Security
               (TLS) Protocol Version 1.2", RFC 5246, August 2008.

   [RFC5944]   Perkins, C., Ed., "IP Mobility Support for IPv4,
               Revised", RFC 5944, November 2010.

16.2.  Informative References



   [CES]       Jen, D., Meisel, M., Yan, H. Massey, D., Wang, L., Zhang,
               B., Zhang, L., "Towards A New Internet Routing
               Architecture: Arguments for Separating Edges from Transit
               Core", 2008, http://conferences.sigcomm.org/
               hotnets/2008/papers/18.pdf.






Frejborg                      Experimental                     [Page 47]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   [Dagstuhl]  Arkko, J., Braun, M.B., Brim, S., Eggert, L., Vogt, C.,
               Zhang, L., "Perspectives Workshop: Naming and Addressing
               in a Future Internet", 2009, http://www.dagstuhl.de/
               de/programm/kalender/semhp/?semnr=09102.

   [ID/loc_Split]
               Thaler, D., "Why do we really want an ID/locator split
               anyway?", 2008,
               http://conferences.sigcomm.org/sigcomm/2008/workshops/
               mobiarch/slides/thaler.pdf.

   [ILNP]      Atkinson, R., "ILNP Concept of Operations", Work in
               Progress, February 2011.

   [iVLB]      Babaioff, M., Chuang, J., "On the Optimality and
               Interconnection of Valiant Load-Balancing Networks",
               2007, http://people.ischool.berkeley.edu/~chuang/
               pubs/VLB-infocom07.pdf.

   [LISP]      Farinacci, D., Fuller, V., Meyer, D., and D. Lewis,
               "Locator/ID Separation Protocol", Work in Progress, June
               2011.

   [Mobility_&_Privacy]
               Brim, S., Linsner. M., McLaughlin, B., and K. Wierenga,
               "Mobility and Privacy", Work in Progress, March 2011.

   [NBS]       Ubillos, J., Xu, M., Ming, Z., and C. Vogt, "Name-Based
               Sockets Architecture", Work in Progress, September 2010.

   [Nimrod]    Chiappa, N., "A New IP Routing and Addressing
               Architecture", 1991, http://ana-3.lcs.mit.edu/
               ~jnc/nimrod/overview.txt.

   [Pathlet_Routing]
               Godfrey, P.G., Shenker, S., Stoica, I., "Pathlet
               Routing", 2008,
               http://conferences.sigcomm.org/hotnets/2008/
               papers/17.pdf.

   [Porting_IPv4]
               DeLong, O., "Porting IPv4 applications to dual stack,
               with examples", 2010,
               http://www.apricot.net/apricot2010/program/tutorials/
               porting-ipv4-apps.html.

   [RBridge]   Perlman, R., "RBridges, Transparent Routing", 2004,
               http://www.ieee-infocom.org/2004/Papers/26_1.PDF.



Frejborg                      Experimental                     [Page 48]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   [Revisiting_Route_Caching]
               Kim, C., Caesar, M., Gerber, A., Rexford, J., "Revisiting
               Route Caching: The World Should Be Flat", 2009,
               http://www.springerlink.com/content/80w13260665v2013/.

   [RFC3597]   Gustafsson, A., "Handling of Unknown DNS Resource Record
               (RR) Types", RFC 3597, September 2003.

   [RFC3618]   Fenner, B., Ed., and D. Meyer, Ed., "Multicast Source
               Discovery Protocol (MSDP)", RFC 3618, October 2003.

   [RFC4423]   Moskowitz, R. and P. Nikander, "Host Identity Protocol
               (HIP) Architecture", RFC 4423, May 2006.

   [RFC4941]   Narten, T., Draves, R., and S. Krishnan, "Privacy
               Extensions for Stateless Address Autoconfiguration in
               IPv6", RFC 4941, September 2007.

   [RFC4960]   Stewart, R., Ed., "Stream Control Transmission Protocol",
               RFC 4960, September 2007.

   [RFC4984]   Meyer, D., Ed., Zhang, L., Ed., and K. Fall, Ed., "Report
               from the IAB Workshop on Routing and Addressing", RFC
               4984, September 2007.

   [RFC5395]   Eastlake 3rd, D., "Domain Name System (DNS) IANA
               Considerations", RFC 5395, November 2008.

   [RFC5880]   Katz, D. and D. Ward, "Bidirectional Forwarding Detection
               (BFD)", RFC 5880, June 2010.

   [RFC6115]   Li, T., Ed., "Recommendation for a Routing Architecture",
               RFC 6115, February 2011.

   [RFC6182]   Ford, A., Raiciu, C., Handley, M., Barre, S., and J.
               Iyengar, "Architectural Guidelines for Multipath TCP
               Development", RFC 6182, March 2011.

   [RFC6227]   Li, T., Ed., "Design Goals for Scalable Internet
               Routing", RFC 6227, May 2011.

   [RRG]       RRG, "IRTF Routing Research Group Home Page",
               http://tools.ietf.org/group/irtf/trac/wiki/
               RoutingResearchGroup.







Frejborg                      Experimental                     [Page 49]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   [SCAFFOLD]  Freedman, M.J., Arye, M., Gopalan, P., Steven Y. Ko,
               S.Y., Nordstrom, E., Rexford, J., Shue, D. "Service-
               Centric Networking with SCAFFOLD", September 2010
               http://www.cs.princeton.edu/research/techreps/TR-885-10.

   [Split-DNS] BIND 9 Administrator Reference Manual,
               http://www.bind9.net/manual/bind/9.3.1/
               Bv9ARM.ch04.html#AEN767.

   [tcpcrypt]  Bittau, A., Hamburg, M., Handley, M., Mazi`eres, D.,
               Boneh, D., "The case for ubiquitous transport-level
               encryption", 2010, http://tcpcrypt.org/tcpcrypt.pdf.

   [VLB]       Zhang-Shen, R., McKeown, N., "Designing a Predictable
               Internet Backbone with Valiant Load-Balancing", 2004,
               http://conferences.sigcomm.org/hotnets/
               2004/HotNets-III%20Proceedings/zhang-shen.pdf.

17.  Acknowledgments



   The active participants at the Routing Research Group [RRG] mailing
   list are acknowledged.  They have provided ideas, proposals, and
   discussions that have influenced the architecture of the hIPv4
   framework.  The following persons, in alphabetical order, have
   provided valuable review input: Aki Anttila, Mohamed Boucadair, Antti
   Jarvenpaa, Dae Young Kim, Mark Lewis, Wes Toman, and Robin Whittle.

   Also, during the IRSG and IESG review process, Rajeev Koodli, Wesley
   Eddy, Jari Arkko, and Adrian Farrel provided valuable review input.

   Lastly, a special thanks to Alfred Schwab from the Poughkeepsie ITSO
   for his editorial assistance.



















Frejborg                      Experimental                     [Page 50]

RFC 6306               Hierarchical IPv4 Framework             July 2011


Appendix A.  Short-Term and Future IPv4 Address Allocation Policy



   In this section, we study how the hIPv4 framework could influence the
   IPv4 address allocation policies to ensure that the new framework
   will enable some reusage of IPv4 address blocks.  It is the Regional
   Internet Registries (RIRs) that shall define the final policies.

   When the intermediate routing architecture (see Figure 1) is fully
   implemented, every ALOC realm could have a full IPv4 address space,
   except the GLB, from which to allocate ELOC blocks.  There are some
   implications, however.  In order for an enterprise to achieve site
   mobility, that is, to change service provider without changing its
   ELOC scheme, the enterprise should implement an autonomous system
   (AS) solution with an ALOC prefix at the attachment point to the
   service provider.

   Larger enterprises have the resources to implement AS border routing.
   Most large enterprises have already implemented multi-homing
   solutions.  Small and midsize enterprises (SMEs) may not have the
   resources to implement AS border routing, or the implementation
   introduces unnecessary costs for the SME.  Also, if every enterprise
   needs to have an allocated ALOC prefix, this will have an impact on
   the RIB at the DFZ -- the RIB will be populated with a huge number of
   non-aggregatable ALOC prefixes.

   It is clear that a compromise is needed.  An SME site usually deploys
   a single uplink to the Internet and should be able to reserve a PI
   ELOC block from the RIR without being forced to create an ALOC realm,
   that is, implement an RBR solution and AS border routing.  Since the
   PI ELOC block is no longer globally unique, an SME can only reserve
   the PI ELOC block for the region where it is active or has its
   attachment point to the Internet.  The attachment point rarely
   changes to another country; therefore, it is sufficient that the PI
   ELOC block is regionally unique.

   When the enterprise replaces its Internet service provider, it does
   not have to change its ELOC scheme -- only the local ALOC prefix at
   the endpoints is changed.  The internal traffic at an enterprise does
   not make use of the ALOC prefix.  The internal routing uses only the
   ELOC prefixes, and thus the internal routing and addressing
   architectures are preserved.

   Mergers and acquisitions of enterprises can cause ELOC conflicts,
   because the PI ELOC block is hereafter only regionally unique.  If an
   enterprise in region A acquires an enterprise in region B, there is a
   slight chance that both enterprises have overlapping ELOC prefixes.





Frejborg                      Experimental                     [Page 51]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   If overlapping of ELOC prefixes occurs, the private unicast ALOC
   solution can be implemented to separate them -- if all affected
   endpoints support the hIPv4 framework.

   Finally, residential users will receive only PA locators.  When a
   residential user changes a service provider, she/he has to replace
   the locators.  Since a PA ELOC block is no longer globally unique,
   every Internet service provider can use the PA ELOC blocks at their
   ALOC realms; the PA locators become kind of private locators for the
   service providers.

   If the forwarding planes and all hosts that establish inter-ALOC
   realm sessions are upgraded to support the hIPv4 framework, that is,
   the long-term routing architecture (see Figure 2) is implemented,
   several interesting possibilities occur:

   o  The regional allocation policy for PI ELOC spaces can be removed,
      and the enterprise can make use of the whole IPv4 address space
      that is globally unique today.  The ELOC space is hereafter only
      significant at a local ALOC realm.

   o  In case of mergers or acquisitions of enterprises, the private
      unicast ALOC solution can be used to separate overlapping ELOC
      spaces.

   o  The GLB space can be expanded to make use of all 32 bits (except
      for the blocks defined in RFC 1918) for anycast and unicast ALOC
      allocations; only ISPs are allowed to apply for GLB prefixes.

   o  The global anycast ALOC solution can be replaced with the global
      unicast ALOC solution since the ISP and enterprise no longer need
      to share ELOC routing information.  Also, there is enough space in
      the GLB to reserve global unicast ALOC prefix(es) for every
      enterprise.

   o  Residential users will still use global anycast ALOC solutions,
      and if they change service providers, their locators need to be
      replaced.

   The result is that a 32x32 bit locator space is achieved.  When an
   enterprise replaces an ISP with another ISP, only the ALOC prefix(es)
   is replaced at endpoints and infrastructure nodes.  Renumbering of
   ALOC prefixes can be automated by, for example, DHCP and extensions
   to IGP.







Frejborg                      Experimental                     [Page 52]

RFC 6306               Hierarchical IPv4 Framework             July 2011


Appendix B.  Multi-Homing becomes Multi-Pathing



   When the transition to the intermediate routing architecture (see
   Figure 1) is fully completed, the RIB of an ISP that has created an
   ALOC realm will have the following entries:

   o  The PA ELOC blocks of directly attached customers (residential and
      enterprises)

   o  The PI ELOC blocks of directly attached customers (e.g.,
      enterprises)

   o  The globally unique ALOC prefixes, received from other service
      providers

   The ISP will not carry any PA or PI ELOC blocks from other service
   providers in its routing table.  In order to do routing and
   forwarding of packets between ISPs, only ALOC information of other
   ISPs is needed.

   Then, the question is how to keep the growth of ALOC reasonable?  If
   the enterprise is using PI addresses, has an AS number, and is
   implementing BGP, why not apply for an ALOC prefix?

   Classical multi-homing is causing the biggest impact on the growth of
   the size of the RIB in the DFZ -- so replacing a /20 IPv4 prefix with
   a /32 ALOC prefix will not reduce the size of the RIB in the DFZ.

   Most likely, the only way to prevent this from happening is to impose
   a yearly cost for the allocation of an ALOC prefix, except if you are
   a service provider that is providing access and/or transit traffic
   for your customers.  And it is reasonable to impose a cost for
   allocating an ALOC prefix for the non-service providers, because when
   an enterprise uses an ALOC prefix, it is reserving a FIB entry
   throughout the DFZ; the ALOC FIB entry needs to have power, space,
   hardware, and cooling on all the routers in the DFZ.

   Implementing this kind of ALOC allocating policy will reduce the RIB
   size in the DFZ quite well, because multi-homing will no longer
   increase the RIB size of the DFZ.  But this policy will have some
   impact on the resilience behavior because by compressing routing
   information we will lose visibility in the network.  In today's
   multi-homing solutions the network always knows where the remote
   endpoint resides.  In case of a link or network failure, a backup
   path is calculated and an alternative path is found, and all routers
   in the DFZ are aware of the change in the topology.  This
   functionality has off-loaded the workload of the endpoints; they only
   need to find the closest ingress router and the network will deliver



Frejborg                      Experimental                     [Page 53]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   the packets to the egress router, regardless (almost) of what
   failures happen in the network.  And with the growth of multi-homed
   prefixes, the routers in the DFZ have been forced to carry greater
   workloads, perhaps close to their limits -- the workload between the
   network and endpoints is not in balance.  The conclusion is that the
   endpoints should take more responsibility for their sessions by
   offloading the workload in the network.  How?  Let us walk through an
   example.

   A remote enterprise has been given an ELOC block 192.168.1.0/24,
   either via static routing or BGP announced to the upstream service
   providers.  The upstream service providers provide the ALOC
   information for the enterprise, 10.1.1.1 and 10.2.2.2.  A remote
   endpoint has been installed and given ELOC 192.168.1.1 -- the ELOC is
   a locator defining where the remote endpoint is attached to the
   remote network.  The remote endpoint has been assigned ALOCs 10.1.1.1
   and 10.2.2.2 -- an ALOC is a locator defining the attachment point of
   the remote network to the Internet.

   The initiator (local endpoint) that has ELOC 172.16.1.1 and ALOC
   prefixes 10.3.3.3 and 10.4.4.4 has established a session by using
   ALOC 10.3.3.3 to the responder (remote endpoint) at ELOC 192.168.1.1
   and ALOC 10.1.1.1.  That is, both networks 192.168.10/24 and
   172.16.1.0/24 are multi-homed.  ALOCs are not available in the
   current IP stack's API, but both ELOCs are seen as the local and
   remote IP addresses in the API, so the application will communicate
   between IP addresses 172.16.1.1 and 192.168.1.1.  If ALOC prefixes
   are included, the session is established between 10.3.3.3:172.16.1.1
   and 10.1.1.1:192.168.1.1.

   Next, a network failure occurs and the link between the responder
   border router (BR-R1) and service provider that owns ALOC 10.1.1.1
   goes down.  The border router of the initiator (BR-I3) will not be
   aware of the situation, because only ALOC information is exchanged
   between service providers and ELOC information is compressed to stay
   within ALOC realms.  But BR-R1 will notice the link failure; BR-R1
   could rewrite the ALOC field in the locator header for this session
   from 10.1.1.1 to 10.2.2.2 and send the packets to the second service
   provider via BR-R2.  The session between the initiator
   10.3.3.3:172.16.1.1 and the responder 10.2.2.2:192.168.1.1 remains
   intact because the legacy 5-tuple at the IP stack API does not
   change.  Only the ALOC prefix of the responder has changed and this
   information is not shown to the application.  An assumption here is
   that the hIPv4 stack does accept changes of ALOC prefixes on the fly
   (more about this later).






Frejborg                      Experimental                     [Page 54]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   If the network link between the BR-I3 and ISP providing ALOC 10.3.3.3
   fails, BR-I3 could rewrite the ALOC prefix in the locator header and
   route the packets via BR-I4 and the session would stay up.  If there
   is a failure somewhere in the network, the border routers might
   receive an ICMP destination unreachable message (if not blocked by
   some security functionality) and thus try to switch the session over
   to the other ISP by replacing the ALOC prefixes in the hIPv4 header.
   Or the endpoints might try themselves to switch to the other ALOCs
   after a certain time-out in the session.  In all session transition
   cases the legacy 5-tuple remains intact.

   If border routers or one of the endpoints changes the ALOC prefix
   without a negotiation with the remote endpoint, security issues
   arise.  Can the endpoints trust the remote endpoint when ALOC
   prefixes are changed on the fly -- is it still the same remote
   endpoint or has the session been hijacked by a bogus endpoint?  The
   obvious answer is that an identification mechanism is needed to
   ensure that after a change in the path or a change of the attachment
   point of the endpoint, the endpoints are still the same.  An
   identifier needs to be exchanged during the transition of the
   session.

   Identifier/locator split schemes have been discussed on the [RRG]
   mailing list, for example, multipath-enabled transport protocols and
   identifier database schemes.  Both types of identifiers can be used
   to protect the session from being hijacked.  A session identifier
   will provide a low-level security mechanism, offering some protection
   against hijacking of the session and also provide mobility.  SCTP
   uses the verification tag to identify the association; MPTCP
   incorporates a token functionality for the same purpose -- both can
   be considered to fulfill the characteristics of a session identifier.
   [tcpcrypt] can be used to further mitigate session hijacking.  If the
   application requires full protection against man-in-the-middle
   attacks, TLS should be applied for the session.  Both transport
   protocols are also multipath-capable.  Implementing multipath-capable
   transport protocols in a multi-homed environment will provide new
   capabilities, such as:

   o  Concurrent and separate exit/entry paths via different attachment
      points at multi-homed sites.

   o  True dynamic load-balancing, in which the endpoints do not
      participate in any routing protocols or do not update rendezvous
      solutions due to network link or node failures.

   o  Only a single Network Interface Card (NIC) on the endpoints is
      required.




Frejborg                      Experimental                     [Page 55]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   o  In case of a border router or ISP failure, the multipath transport
      protocol will provide resilience.

   By adding more intelligence at the endpoints, such as multipath-
   enabled transport protocols, the workload of the network is offloaded
   and can take less responsibility for providing visibility of
   destination prefixes on the Internet; for example, prefix compression
   in the DFZ can be applied and only the attachment points of a local
   network need to be announced in the DFZ.  And the IP address space no
   longer needs to be globally unique; it is sufficient that only a part
   is globally unique, with the rest being only regionally unique (in
   the long-term routing architecture, locally unique) as discussed in
   Appendix A.

   The outcome is that the current multi-homing solution can migrate
   towards a multi-pathing environment that will have the following
   characteristics:

   o  An AS number is not mandatory for enterprises.

   o  BGP is not mandatory at the enterprise's border routers; static
      routing with Bidirectional Forwarding Detection (BFD) [RFC5880] is
      an option.

   o  Allocation of global ALOC prefixes for the enterprise should not
      be allowed; instead, upstream ISPs provide the global ALOC
      prefixes for the enterprise.

   o  MPTCP provides dynamic load-balancing without using routing
      protocols; several paths can be used simultaneously and thus
      resilience is achieved.

   o  Provides low growth of RIB entries at the DFZ.

   o  When static routing is used between the enterprise and the ISP:

      -  The RIB size at the enterprise's border routers does not depend
         upon the size of the RIB in the DFZ or in adjacent ISPs.

      -  The enterprise's border router cannot cause BGP churn in the
         DFZ or in the adjacent ISPs' RIB.

   o  When dynamic routing is used between the enterprise and the ISP:

      -  The RIB size at the enterprise's border routers depends upon
         the size of the RIB in the DFZ and adjacent ISPs.





Frejborg                      Experimental                     [Page 56]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      -  The enterprise's border router can cause BGP churn for the
         adjacent ISPs, but not in the DFZ.

   o  The cost of the border router should be less than in today's
      multi-homing solution.

Appendix C.  Incentives and Transition Arguments



   The media has announced the meltdown of the Internet and the
   depletion of IPv4 addresses several times, but the potential chaos
   has been postponed and the general public has lost interest in these
   announcements.  Perhaps it could be worthwhile to find other valuable
   arguments that the general public could be interested in, such as:

   o  Not all endpoints need to be upgraded, only those that are
      directly attached to the Internet, such as portable laptops, smart
      mobile phones, proxies, and DMZ/frontend endpoints.  But the most
      critical endpoints, the backend endpoints where enterprises keep
      their most critical business applications, do not need to be
      upgraded.  These endpoints should not be reached at all from the
      Internet, only from the private network.  And this functionality
      can be achieved with the hIPv4 framework, since it is backwards
      compatible with the current IPv4 stack.  Therefore, investments in
      legacy applications used inside an ALOC realm are preserved.

   o  Mobility - it is estimated that the demand for applications that
      perform well over the wireless access network will increase.
      Introduction of MPTCP and identifier/locator split schemes opens
      up new possibilities to create new solutions and applications that
      are optimized for mobility.  The hIPv4 framework requires an
      upgrade of the endpoint's stack; if possible, the hIPv4 stack
      should also contain MPTCP and identifier/locator split scheme
      features.  Applications designed for mobility could bring
      competitive benefits.

   o  The intermediate routers in the network do not need to be upgraded
      immediately; the current forwarding plane can still be used.  The
      benefit is that the current network equipment can be preserved at
      the service providers, enterprises, and residences (except
      middleboxes).  This means that the carbon footprint is a lot lower
      compared to other solutions.  Many enterprises do have green
      programs and many residential users are concerned with the global
      warming issue.

   o  The migration from IPv4 to IPv6 (currently defined architecture)
      will increase the RIB and FIB throughout DFZ.  Whether it will
      require a new upgrade of the forwarding plane as discussed in
      [RFC4984] is unclear.  Most likely an upgrade is needed.  The



Frejborg                      Experimental                     [Page 57]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      outcome of deploying IPv4 and IPv6 concurrently is that the
      routers need to have larger memories for the RIB and FIB -- every
      globally unique prefix is installed in the routers that are
      participating in the DFZ.  Since the enterprise reserves one or
      several RIB/FIB entries on every router in the DFZ, it is
      increasing the power consumption of the Internet, thus increasing
      the carbon footprint.  And many enterprises are committed to green
      programs.  If hIPv4 is deployed, the power consumption of the
      Internet will not grow as much as in an IPv4 to IPv6 transition
      scenario.

   o  Another issue: if the migration from IPv4 to IPv6 (currently
      defined architecture) occurs, the routers in the DFZ most likely
      need to be upgraded to more expensive routers, as discussed in
      [RFC4984].  In the wealthy part of the world, where a large
      penetration of Internet users is already present, the service
      providers can pass the costs of the upgrade along to their
      subscribers more easily.  With a "wealthy/high penetration" ratio
      the cost will not grow so much that the subscribers would abandon
      the Internet.  But in the less wealthy part of the world, where
      there is usually a lower penetration of subscribers, the cost of
      the upgrade cannot be accepted so easily -- a "less wealthy/low
      penetration" ratio could impose a dramatic increase of the cost
      that needs to be passed along to the subscribers.  And thus fewer
      subscribers could afford to get connected to the Internet.  For
      the global enterprises and the enterprises in the less wealthy
      part of the world, this scenario could mean less potential
      customers and there could be situations when the nomads of the
      enterprises can't get connected to the Internet.  This is also not
      fair; every human being should have a fair chance to be able to
      enjoy the Internet experience -- and the wealthy part of the world
      should take this right into consideration.  Many enterprises are
      committed to Corporate Social Responsibility programs.

   Not only technical and economical arguments can be found.  Other
   arguments that the general public is interested in and concerned
   about can be found, for example, that the Internet becomes greener
   and more affordable for everyone, in contrast with the current
   forecast of the evolution of the Internet.

Appendix D.  Integration with CES Architectures



   Because the hIPv4 framework requires changes to the endpoint's stack,
   it will take some time before the migration of the current IPv4
   architecture to the intermediate hIPv4 routing architecture is fully
   completed.  If a hIPv4 proxy solution could be used in front of





Frejborg                      Experimental                     [Page 58]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   classical IPv4 endpoints, the threshold for early adopters to start
   to migrate towards the hIPv4 framework would be less questionable and
   the migration phase would also most likely be much shorter.

   Therefore, it should be investigated whether the hIPv4 framework can
   be integrated with Core-Edge Separation [CES] architectures.  In CES
   architectures the endpoints do not need to be modified.  The design
   goal of a CES solution is to minimize the PI-address entries in the
   DFZ and to preserve the current stack at the endpoints.  But a CES
   solution requires a new mapping system and also introduces a caching
   mechanism in the map-and-encapsulate network nodes.  Much debate
   about scalability of a mapping system and the caching mechanism has
   been going on at the [RRG] list.  At the present time it is unclear
   how well both solutions will scale; research work on both topics is
   still in progress.

   Since the CES architectures divide the address spaces into two new
   categories, one that is installed in the RIB of the DFZ and one that
   is installed in the local networks, there are to some degree
   similarities between CES architectures and the hIPv4 framework.
   Actually, the invention of the IP and locator header swap
   functionality was inspired by [LISP].

   In order to describe how these two architectures might be integrated,
   some terminology definitions are needed:

   CES-node:

      A network node installed in front of a local network that must
      have the following characteristics:

         o  Map-and-encapsulate ingress functionality

         o  Map-and-encapsulate egress functionality

         o  Incorporate the hIPv4 stack

         o  Routing functionality, [RFC1812]

         o  Being able to apply policy-based routing on the ALOC field
            in the locator header

      The CES-node does not include the MPTCP extension because it would
      most likely put too much of a burden on the CES-node to signal and
      maintain MPTCP subflows for the cached hIPv4 entries.






Frejborg                      Experimental                     [Page 59]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   Consumer site:

      A site that is not publishing any services towards the Internet,
      that is, there are no entries in DNS for this site.  It is used by
      local endpoints to establish outbound connectivity -- endpoints
      are initiating sessions from the site towards content sites.
      Usually such sites are found at small enterprises and residences.
      PA-addresses are usually assigned to them.

   Content site:

      A site that is publishing services towards the Internet, and which
      usually does have DNS entries.  Such a site is used by local
      endpoints to establish both inbound and outbound connectivity.
      Large enterprises use PI-addresses, while midsize/small
      enterprises use either PI- or PA-address space.

   The CES architectures aim to reduce the PI-address entries in the
   DFZ.  Therefore, map-and-encapsulate egress functionality will be
   installed in front of the content sites.  It is likely that the node
   containing map-and-encapsulate egress functionality will also contain
   map-and-encapsulate ingress functionality; it is also a router, so
   the node just needs to support the hIPv4 stack and be able to apply
   policy-based routing using the ALOC field of the locator header to
   become a CES-node.

   It is possible that the large content providers (LCPs) are not
   willing to install map-and-encapsulate functionality in front of
   their sites.  If the caching mechanism is not fully reliable or if
   the mapping lookup delay does have an impact on their clients' user
   experience, then most likely the LCPs will not adopt the CES
   architecture.

   In order to convince a LCP to adopt the CES architecture, it should
   provide a mechanism to mitigate the caching and mapping lookup delay
   risks.  One method is to push the CES architectures to the edge --
   the closer to the edge you add new functionality, the better it will
   scale.  That is, if the endpoint stack is upgraded, the caching
   mechanism is maintained by the endpoint itself.  The mapping
   mechanism can be removed if the CES architecture's addressing scheme
   is replaced with the addressing scheme of hIPv4 when the CES solution
   is integrated at the endpoints.  With this approach, the LCPs might
   install a CES-node in front of their sites.  Also, some endpoints at
   the content site might be upgraded with the hIPv4 stack.







Frejborg                      Experimental                     [Page 60]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   If the LCP faces issues with the caching or mapping mechanisms, the
   provider can ask its clients to upgrade their endpoint's stack to
   ensure a proper service level.  At the same time, the LCP promotes
   the migration from the current routing architecture to a new routing
   architecture, not for the sake of the routing architecture but
   instead to ensure a proper service level -- you can say that a
   business model will promote the migration of a new routing
   architecture.

   The hIPv4 framework proposes that the IPv4 addresses (ELOC) should no
   longer be globally unique; once the transition is completed, a more
   regional allocation can be deployed.  But this is only possible once
   all endpoints (that are establishing sessions to other ALOC realms)
   have migrated to support the hIPv4 framework.  Here the CES
   architecture can speed up the re-usage of IPv4 addresses; that is,
   once an IPv4 address block has become an ELOC block it can be re-used
   in the other RIR regions, without the requirement that all endpoints
   in the Internet must first be upgraded.

   As stated earlier, the CES architecture aims to remove PI-addresses
   from the DFZ, making the content sites more or less the primary
   target for the roll-out of a CES solution.  At large content sites a
   CES-node most likely will be installed.  To upgrade all endpoints
   (that are providing services towards the Internet) at a large content
   site will take time, and it might be that the endpoints at the
   content site are upgraded only within their normal lifecycle process.
   But if the size of the content site is small, the administrator
   either installs a CES-node or upgrades the endpoint's stack -- a
   decision influenced by availability, reliability, and economic
   feasibility.

   Once the content sites have been upgraded, the PI-address entries
   have been removed from the DFZ.  Most likely also some endpoints at
   the consumer sites have been upgraded to support the hIPv4 stack --
   especially if there have been issues with the caches or mapping
   delays that have influenced the service levels at the LCPs.  Then,
   the issue is how to keep track of the upgrade of the content sites --
   have they been migrated or not?  If the content sites or content
   endpoints have been migrated, the DNS records should have either a
   CES-node entry or ALOC entry for each A-record.  When the penetration
   of CES solutions at content sites (followed up by CES-node/ALOC
   records in DNS) is high enough, the ISP can start to promote the
   hIPv4 stack upgrade at the consumer sites.

   Once a PA-address block has been migrated it can be released from
   global allocation to a regional allocation.  Why would an ISP then
   push its customers to deploy hIPv4 stacks?  Because of the business
   model -- it will be more expensive to stay in the current



Frejborg                      Experimental                     [Page 61]

RFC 6306               Hierarchical IPv4 Framework             July 2011


   architecture.  The depletion of IPv4 addresses will either cause more
   NAT at the service provider's network (operational expenditures will
   increase because the network will become more complex) or the ISP
   should force its customers to migrate to IPv6.  But the ISP could
   lose customers to other ISPs that are offering IPv4 services.

   When PA-addresses have been migrated to the hIPv4 framework, the ISP
   will have a more independent routing domain (ALOC realm) with only
   ALOC prefixes from other ISPs and ELOC prefixes from directly
   attached customers.  BGP churn from other ISPs is no longer received,
   the amount of alternative paths is reduced, and the ISP can better
   control the growth of the RIB at their ALOC realm.  The operational
   and capital expenditures should be lower than in the current routing
   architecture.

   To summarize, the content providers might find the CES+hIPv4 solution
   attractive.  It will remove the forthcoming IPv4 address depletion
   constraints without forcing the consumers to switch to IPv6, and thus
   the content providers can continue to grow (reach more consumers).

   The ISP might also find this solution attractive because it should
   reduce the capital and operational expenditures in the long term.
   Both the content providers and the ISPs are providing the foundation
   of the Internet.  If both adopt this architecture, the consumers have
   to adopt.  Both providers might find business models to "guide" the
   consumers towards the new routing architecture.

   Then, how will this affect the consumer and content sites?
   Residential users will need to upgrade their endpoints.  But it
   doesn't really matter which IP version they use.  It is the
   availability and affordability of the Internet that matters most.

   Enterprises will be affected a little bit more.  The edge devices at
   the enterprises' local networks need to be upgraded -- edge nodes
   such as AS border routers, middleboxes, DNS, DHCP, and public nodes
   -- but by installing a CES-node in front of them, the upgrade process
   is postponed and the legacy nodes can be upgraded during their normal
   lifecycle process.  The internal infrastructure is preserved,
   internal applications can still use IPv4, and all investment in IPv4
   skills is preserved.

   Walkthrough of use cases:

   1. A legacy endpoint at a content site establishes a session to a
      content site with a hIPv4 upgraded endpoint.






Frejborg                      Experimental                     [Page 62]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      When the legacy endpoint resolves the DNS entry for the remote
      endpoint (a hIPv4 upgraded endpoint), it receives an ALOC record
      in the DNS response.  The legacy endpoint ignores the ALOC record.
      Only the A-record is used to establish the session.  Next, the
      legacy endpoint initializes the session and a packet is sent
      towards the map-and-encapsulate ingress node, which needs to do a
      lookup at the CES mapping system (the assumption here is that no
      cache entry exists for the remote endpoint).  The mapping system
      returns either a CES-node prefix or an ALOC prefix for the lookup
      -- since the requested remote endpoint has been upgraded, the
      mapping system returns an ALOC prefix.

      The CES-node will not use the CES encapsulation scheme for this
      session.  Instead, the hIPv4 header scheme will be used and a /32
      entry will be created in the cache.  A /32 entry must be created;
      it is possible that not all endpoints at the remote site are
      upgraded to support the hIPv4 framework.  The /32 cache entry can
      be replaced with a shorter prefix in the cache if all endpoints
      are upgraded at the remote site.  To indicate this situation, a
      subfield should be added for the ALOC record in the mapping
      system.

      The CES-node must execute the following steps for the egress
      packets:

      a. Verify IP and transport header checksums.

      b. Create the locator header and copy the value in the destination
         address field of the IP header to the ELOC field of the locator
         header.

      c. Replace the destination address in the IP header with the ALOC
         prefix given in the cache.

      d. Insert the local CES-node prefix in the ALOC field of the
         locator header.

      e. Copy the transport protocol value of the IP header to the
         protocol field of the locator header and set the hIPv4 protocol
         value in the protocol field of the IP header.

      f. Set the desired parameters in the A-, I-, S-, VLB-, and L-
         fields of the locator header.

      g. Set the FI-bits of the locator header to 00.

      h. Decrease the TTL value by one.




Frejborg                      Experimental                     [Page 63]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      i. Calculate IP, locator, and transport protocol header checksums.
         Transport protocol header calculations do not include the
         locator header fields.  When completed, the packet is
         transmitted.

      j. Because the size of the packet might exceed MTU due to the
         insertion of the locator header, and if MTU is exceeded, the
         CES-node should inform the source endpoint of the situation
         with an ICMP message, and the CES-node should apply
         fragmentation of the hIPv4 packet.

   2. A hIPv4-upgraded endpoint at a consumer/content site establishes a
      session to a content site with a CES-node in front of a legacy
      endpoint.

      The hIPv4 upgraded endpoint receives, in the DNS response, either
      an ALOC record or a CES-node record for the resolved destination.
      From the requesting hIPv4 endpoint's point of view, it really
      doesn't matter if the new record prefix is used to locate RBR-
      nodes or CES-nodes in the Internet -- the CES-node will act as a
      hIPv4 proxy in front of the remote legacy endpoint.  Thus the
      hIPv4 endpoint assembles a hIPv4 packet to initialize the session,
      and when the packet arrives at the CES-node it must execute the
      following:

      a. Verify that the received packet uses the hIPv4 protocol value
         in the protocol field of the IP header.

      b. Verify IP, locator, and transport protocol header checksums.
         Transport protocol header verification does not include the
         locator header fields.

      c. Replace the protocol field value of the IP header with the
         protocol field value of the locator header.

      d. Replace the destination address in the IP header with the ELOC
         prefix of the locator header.

      e. Remove the locator header.

      f. Create a cache entry (unless an entry already exists) for
         returning packets.  A /32 entry is required.  To optimize the
         usage of cache entries, the CES-node might ask the CES mapping
         node whether all endpoints at the remote site are upgraded or
         not.  If upgraded, a shorter prefix can be used in the cache.

      g. Decrease the TTL value by one.




Frejborg                      Experimental                     [Page 64]

RFC 6306               Hierarchical IPv4 Framework             July 2011


      h. Calculate IP and transport protocol header checksums.

      i. Forward the packet according to the destination address of the
         IP header.

   3. A hIPv4-enabled endpoint with a regionally unique ELOC at a
      consumer site establishes a session to a consumer site with a
      legacy endpoint.

      In this use case, the sessions will fail unless some mechanism is
      invented and implemented at the ISPs' map-and-encapsulate nodes.
      The sessions will work inside an ALOC realm since the classical
      IPv4 framework is still valid.  Sessions between ALOC realms will
      fail.  Some applications establish sessions between consumer
      sites.  The most common are gaming and peer-to-peer applications.
      These communities have historically been in the forefront of
      adopting new technologies.  It is expected that they either
      develop workarounds to solve this issue or simply ask their
      members to upgrade their stacks.

   4. A legacy endpoint at a consumer/content site establishes a session
      to a content site with a CES-node in front of a legacy endpoint.

      Assumed to be described in CES architecture documents.

   5. A hIPv4-enabled endpoint at a consumer/content site establishes a
      session to a content site with a hIPv4-enabled endpoint.

      See Section 5.2.

Author's Address



   Patrick Frejborg
   EMail: pfrejborg@gmail.com

















Frejborg                      Experimental                     [Page 65]