RFC 7018






Internet Engineering Task Force (IETF)                         V. Manral
Request for Comments: 7018                                            HP
Category: Informational                                         S. Hanna
ISSN: 2070-1721                                                  Juniper
                                                          September 2013


         Auto-Discovery VPN Problem Statement and Requirements

Abstract



   This document describes the problem of enabling a large number of
   systems to communicate directly using IPsec to protect the traffic
   between them.  It then expands on the requirements for such a
   solution.

   Manual configuration of all possible tunnels is too cumbersome in
   many such cases.  In other cases, the IP addresses of endpoints
   change, or the endpoints may be behind NAT gateways, making static
   configuration impossible.  The Auto-Discovery VPN solution will
   address these requirements.

Status of This Memo



   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Not all documents
   approved by the IESG are 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/rfc7018.














Manral & Hanna                Informational                     [Page 1]

RFC 7018                   Auto-Discovery VPN             September 2013


Copyright Notice



   Copyright (c) 2013 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.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents



   1. Introduction ....................................................2
      1.1. Terminology ................................................3
      1.2. Conventions Used in This Document ..........................4
   2. Use Cases .......................................................4
      2.1. Use Case 1: Endpoint-to-Endpoint VPN .......................4
      2.2. Use Case 2: Gateway-to-Gateway VPN .........................5
      2.3. Use Case 3: Endpoint-to-Gateway VPN ........................6
   3. Inadequacy of Existing Solutions ................................6
      3.1. Exhaustive Configuration ...................................6
      3.2. Star Topology ..............................................6
      3.3. Proprietary Approaches .....................................7
   4. Requirements ....................................................7
      4.1. Gateway and Endpoint Requirements ..........................7
   5. Security Considerations ........................................11
   6. Acknowledgements ...............................................11
   7. Normative References ...........................................12

1.  Introduction



   IPsec [RFC4301] is used in several different cases, including
   tunnel-mode site-to-site VPNs and remote access VPNs.  Both tunneling
   modes for IPsec gateways and host-to-host transport mode are
   supported in this document.

   The subject of this document is the problem presented by large-scale
   deployments of IPsec and the requirements on a solution to address
   the problem.  These may be a large collection of VPN gateways
   connecting various sites, a large number of remote endpoints
   connecting to a number of gateways or to each other, or a mix of the
   two.  The gateways and endpoints may belong to a single
   administrative domain or several domains with a trust relationship.



Manral & Hanna                Informational                     [Page 2]

RFC 7018                   Auto-Discovery VPN             September 2013


   Section 4.4 of RFC 4301 describes the major IPsec databases needed
   for IPsec processing.  It requires extensive configuration for each
   tunnel, so manually configuring a system of many gateways and
   endpoints becomes infeasible and inflexible.

   The difficulty is that a lot of configuration mentioned in RFC 4301
   is required to set up a Security Association.  The Internet Key
   Exchange Protocol (IKE) implementations need to know the identity and
   credentials of all possible peer systems, as well as the addresses of
   hosts and/or networks behind them.  A simplified mechanism for
   dynamically establishing point-to-point tunnels is needed.  Section 2
   contains several use cases that motivate this effort.

1.1.  Terminology



   Auto-Discovery Virtual Private Network (ADVPN) -  A VPN solution that
      enables a large number of systems to communicate directly, with
      minimal configuration and operator intervention, using IPsec to
      protect communication between them.

   Endpoint -  A device that implements IPsec for its own traffic but
      does not act as a gateway.

   Gateway -  A network device that implements IPsec to protect traffic
      flowing through the device.

   Point-to-Point -  Communication between two parties without active
      participation (e.g., encryption or decryption) by any other
      parties.

   Hub -  The central point in a star topology/dynamic full-mesh
      topology, or one of the central points in the full-mesh style VPN,
      i.e., a gateway to which multiple other hubs or spokes connect.
      The hubs usually forward traffic coming from encrypted links to
      other encrypted links, i.e., there are no devices connected to
      them in the clear.

   Spoke -  The endpoint in a star topology/dynamic full-mesh topology
      or gateway that forwards traffic from multiple cleartext devices
      to other hubs or spokes, and some of those other devices are
      connected to it in the clear (i.e., it encrypts data coming from
      cleartext devices and forwards it to the ADVPN).

   ADVPN Peer -  Any member of an ADVPN, including gateways, endpoints,
      hubs, or spokes.






Manral & Hanna                Informational                     [Page 3]

RFC 7018                   Auto-Discovery VPN             September 2013


   Star Topology -  Topology in which there is direct connectivity only
      between the hub and spoke, and where communication between the 2
      spokes happens through the hub.

   Allied and Federated Environments -  Environments where we have
      multiple different organizations that have close associations and
      need to connect to each other.

   Full-Mesh Topology -  Topology in which there is direct connectivity
      between every spoke to every other spoke, without the traffic
      between the spokes having to be redirected through an intermediate
      hub device.

   Dynamic Full-Mesh Topology -  Topology in which direct connections
      exist in a hub-and-spoke manner but dynamic connections are
      created/removed between the spokes on an as-needed basis.

   Security Association (SA) -  Defined in [RFC4301].

1.2.  Conventions Used in This Document



   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].

2.  Use Cases



   This section presents the key use cases for large-scale
   point-to-point VPNs.

   In all of these use cases, the participants (endpoints and gateways)
   may be from a single organization (administrative domain) or from
   multiple organizations with an established trust relationship.  When
   multiple organizations are involved, products from multiple vendors
   are employed, so open standards are needed to provide
   interoperability.  Establishing communications between participants
   with no established trust relationship is out of scope for this
   effort.

2.1.  Use Case 1: Endpoint-to-Endpoint VPN



   Two endpoints wish to communicate securely via a point-to-point SA.

   The need for secure endpoint-to-endpoint communications is often
   driven by a need to employ high-bandwidth, low-latency local
   connectivity instead of using slow, expensive links to remote
   gateways.  For example, two users in close proximity may wish to
   place a direct, secure video or voice call without needing to send



Manral & Hanna                Informational                     [Page 4]

RFC 7018                   Auto-Discovery VPN             September 2013


   the call through remote gateways, as the remote gateways would add
   latency to the call, consume precious remote bandwidth, and increase
   overall costs.  Such a use case also enables connectivity when both
   users are behind NAT gateways.  Such a use case ought to allow for
   seamless connectivity even as endpoints roam and even if they are
   moving out from behind a NAT gateway, from behind one NAT gateway to
   behind another, or from a standalone position to behind a NAT
   gateway.

   In a star topology, when two endpoints communicate, they need a
   mechanism for authentication such that they do not expose themselves
   to impersonation by the other spoke endpoint.

2.2.  Use Case 2: Gateway-to-Gateway VPN



   A typical Enterprise traffic model follows a star topology, with the
   gateways connecting to each other using IPsec tunnels.

   However, for voice and other rich media traffic that require a lot of
   bandwidth or is performance sensitive, the traffic tromboning (taking
   a suboptimal path) to the hub can create traffic bottlenecks on the
   hub and can lead to an increase in cost.  A fully meshed solution
   would make best use of the available network capacity and
   performance, but the deployment of a fully meshed solution involves
   considerable configuration, especially when a large number of nodes
   are involved.  It is for this purpose that spoke-to-spoke tunnels are
   dynamically created and torn down.  For the reasons of cost and
   manual error reduction, it is desired that there be minimal
   configuration on each gateway.

   The solution ought to work in cases where the endpoints are in
   different administrative domains that have an existing trust
   relationship (for example, two organizations that are collaborating
   on a project may wish to join their networks while retaining
   independent control over configuration).  It is highly desirable that
   the solution works for the star, full-mesh, and dynamic full-mesh
   topologies.

   The solution ought to also address the case where gateways use
   dynamic IP addresses.

   Additionally, the routing implications of gateway-to-gateway
   communication need to be addressed.  In the simple case, selectors
   provide sufficient information for a gateway to forward traffic
   appropriately.  In other cases, additional tunneling (e.g., Generic
   Routing Encapsulation (GRE)) and routing (e.g., Open Shortest Path
   First (OSPF)) protocols are run over IPsec tunnels, and the
   configuration impact on those protocols needs to be considered.



Manral & Hanna                Informational                     [Page 5]

RFC 7018                   Auto-Discovery VPN             September 2013


   There is also the case where Layer 3 Virtual Private Networks
   (L3VPNs) operate over IPsec tunnels.

   When two gateways communicate, they need to use a mechanism for
   authentication such that they do not expose themselves to the risk of
   impersonation by the other entities.

2.3.  Use Case 3: Endpoint-to-Gateway VPN



   A mobile endpoint ought to be able to use the most efficient gateway
   as it roams in the Internet.

   A mobile user roaming on the Internet may connect to a gateway that,
   because of roaming, is no longer the most efficient gateway to use
   (reasons could be cost, efficiency, latency, or some other factor).
   The mobile user ought to be able to discover and then connect to the
   current, most efficient gateway in a seamless way without having to
   bring down the connection.

3.  Inadequacy of Existing Solutions



   Several solutions exist for the problems described above.  However,
   none of these solutions is adequate, as described here.

3.1.  Exhaustive Configuration



   One simple solution is to configure all gateways and endpoints in
   advance with all the information needed to determine which gateway or
   endpoint is optimal and to establish an SA with that gateway or
   endpoint.  However, this solution does not scale in a large network
   with hundreds of thousands of gateways and endpoints, especially when
   multiple administrative domains are involved and things are rapidly
   changing (e.g., mobile endpoints).  Such a solution is also limited
   by the smallest endpoint/gateway, as the same exhaustive
   configuration is to be applied on all endpoints/gateways.  A more
   dynamic, secure, and scalable system for establishing SAs between
   gateways is needed.

3.2.  Star Topology



   The most common way to address a part of this problem today is to use
   what has been termed a "star topology".  In this case, one or a few
   gateways are defined as "hub gateways", while the rest of the systems
   (whether endpoints or gateways) are defined as "spokes".  The spokes
   never connect to other spokes.  They only open tunnels with the hub
   gateways.  Also, for a large number of gateways in one administrative
   domain, one gateway may be defined as the hub, and the rest of the
   gateways and remote access clients connect only to that gateway.



Manral & Hanna                Informational                     [Page 6]

RFC 7018                   Auto-Discovery VPN             September 2013


   This solution, however, is complicated by the case where the spokes
   use dynamic IP addresses and DNS with dynamic updates needs to be
   used.  It is also desired that there is minimal to no configuration
   on the hub as the number of spokes increases and new spokes are added
   and deleted randomly.

   Another problem with the star topology is that it creates a high load
   on the hub gateways, as well as on the connection between the spokes
   and the hub.  This load impacts both processing power and network
   bandwidth.  A single packet in the hub-and-spoke scenario can be
   encrypted and decrypted multiple times.  It would be much preferable
   if these gateways and clients could initiate tunnels between them,
   bypassing the hub gateways.  Additionally, the path bandwidth to
   these hub gateways may be lower than that of the path between the
   spokes.  For example, two remote access users may be in the same
   building with high-speed WiFi (for example, at an IETF meeting).
   Channeling their conversation through the hub gateways of their
   respective employers seems extremely wasteful, given that a more
   optimal direct path exists.

   The challenge is to build large-scale IPsec-protected networks that
   can dynamically change with minimal administrative overhead.

3.3.  Proprietary Approaches



   Several vendors offer proprietary solutions to these problems.
   However, these solutions offer no interoperability between equipment
   from one vendor and another.  This means that they are generally
   restricted to use within one organization, and it is harder to move
   away from such solutions, as the features are not standardized.
   Besides, multiple organizations cannot be expected to all choose the
   same equipment vendor.

4.  Requirements



   This section defines the requirements on which the solution will be
   based.

4.1.  Gateway and Endpoint Requirements



   1.   For any network topology (star, full mesh, and dynamic full
        mesh), when a new gateway or endpoint is added, removed, or
        changed, configuration changes are minimized as follows.  Adding
        or removing a spoke in the topology MUST NOT require
        configuration changes to hubs other than where the spoke was
        connected and SHOULD NOT require configuration changes to the
        hub to which the spoke was connected.  The changes also MUST NOT
        require configuration changes in other spokes.



Manral & Hanna                Informational                     [Page 7]

RFC 7018                   Auto-Discovery VPN             September 2013


        Specifically, when evaluating potential proposals, we will
        compare them by looking at how many endpoints or gateways must
        be reconfigured when a new gateway or endpoint is added,
        removed, or changed and how substantial this reconfiguration is,
        in addition to the amount of static configuration required.

        This requirement is driven by use cases 1 and 2 and by the
        scaling limitations pointed out in Section 3.1.

   2.   ADVPN Peers MUST allow IPsec tunnels to be set up with other
        members of the ADVPN without any configuration changes, even
        when peer addresses get updated every time the device comes up.
        This implies that Security Policy Database (SPD) entries or
        other configuration based on a peer IP address will need to be
        automatically updated, avoided, or handled in some manner to
        avoid a need to manually update policy whenever an address
        changes.

   3.   In many cases, additional tunneling protocols (e.g., GRE) or
        routing protocols (e.g., OSPF) are run over the IPsec tunnels.
        Gateways MUST allow for the operation of tunneling and routing
        protocols operating over spoke-to-spoke IPsec tunnels with
        minimal or no configuration impact.  The ADVPN solution SHOULD
        NOT
increase the amount of information required to configure
        protocols running over IPsec tunnels.

   4.   In the full-mesh and dynamic full-mesh topologies, spokes MUST
        allow for direct communication with other spoke gateways and
        endpoints.  In the star topology mode, direct communication
        between spokes MUST be disallowed.

        This requirement is driven by use cases 1 and 2 and by the
        limitations of a star topology as pointed out in Section 3.2.

   5.   ADVPN Peers MUST NOT have a way to get the long-term
        authentication credentials for any other ADVPN Peers.  The
        compromise of an endpoint MUST NOT affect the security of
        communications between other ADVPN Peers.  The compromise of a
        gateway SHOULD NOT affect the security of the communications
        between ADVPN Peers not associated with that gateway.

        This requirement is driven by use case 1.  ADVPN Peers
        (especially spokes) become compromised fairly often.  The
        compromise of one ADVPN Peer SHOULD NOT affect the security of
        other unrelated ADVPN Peers.






Manral & Hanna                Informational                     [Page 8]

RFC 7018                   Auto-Discovery VPN             September 2013


   6.   Gateways SHOULD allow for seamless handoff of sessions in cases
        where endpoints are roaming, even if they cross policy
        boundaries.  This would mean the data traffic is minimally
        affected even as the handoff happens.  External factors like
        firewalls and NAT boxes that will be part of the overall
        solution when ADVPN is deployed will not be considered part of
        this solution.

        Such endpoint roaming may affect not only the endpoint-to-
        endpoint SA but also the relationship between the endpoints and
        gateways (such as when an endpoint roams to a new network that
        is handled by a different gateway).

        This requirement is driven by use case 1.  Today's endpoints are
        mobile and transition often between different networks (from 4G
        to WiFi and among various WiFi networks).

   7.   Gateways SHOULD allow for easy handoff of a session to another
        gateway, to optimize latency, bandwidth, load balancing,
        availability, or other factors, based on policy.

        This ability to migrate traffic from one gateway to another
        applies regardless of whether the gateways in question are hubs
        or spokes.  It even applies in the case where a gateway (hub or
        spoke) moves in the network, as may happen with a vehicle-based
        network.

        This requirement is driven by use case 3.

   8.   Gateways and endpoints MUST have the capability to participate
        in an ADVPN even when they are located behind NAT boxes.
        However, in some cases they may be deployed in such a way that
        they will not be fully reachable behind a NAT box.  It is
        especially difficult to handle cases where the hub is behind a
        NAT box.  When the two endpoints are both behind separate NATs,
        communication between these spokes SHOULD be supported using
        workarounds such as port forwarding by the NAT or detecting when
        two spokes are behind uncooperative NATs, and using a hub in
        that case.

        This requirement is driven by use cases 1 and 2.  Endpoints are
        often behind NATs, and gateways sometimes are.  IPsec SHOULD
        continue to work seamlessly regardless, using ADVPN techniques
        whenever possible and providing graceful fallback to hub-and-
        spoke techniques as needed.






Manral & Hanna                Informational                     [Page 9]

RFC 7018                   Auto-Discovery VPN             September 2013


   9.   Changes such as establishing a new IPsec SA SHOULD be reportable
        and manageable.  However, creating a MIB or other management
        technique is not within scope for this effort.

        This requirement is driven by manageability concerns for all the
        use cases, especially use case 2.  As IPsec networks become more
        dynamic, management tools become more essential.

   10.  To support allied and federated environments, endpoints and
        gateways from different organizations SHOULD be able to connect
        to each other.

        This requirement is driven by demand for all the use cases in
        federated and allied environments.

   11.  The administrator of the ADVPN SHOULD allow for the
        configuration of a star, full-mesh, or partial full-mesh
        topology, based on which tunnels are allowed to be set up.

        This requirement is driven by demand for all the use cases in
        federated and allied environments.

   12.  The ADVPN solution SHOULD be able to scale for multicast
        traffic.



        This requirement is driven by use case 2, where the amount of
        rich media multicast traffic is increasing.

   13.  The ADVPN solution SHOULD allow for easy monitoring, logging,
        and reporting of the dynamic changes to help with
        troubleshooting such environments.

        This requirement is driven by demand for all the use cases in
        federated and allied environments.

   14.  There is also the case where L3VPNs operate over IPsec tunnels,
        for example, Provider-Edge-based VPNs.  An ADVPN MUST support
        L3VPNs as applications protected by the IPsec tunnels.

        This requirement is driven by demand for all the use cases in
        federated and allied environments.

   15.  The ADVPN solution SHOULD allow the enforcement of per-peer QoS
        in both the star and full-mesh topologies.



        This requirement is driven by demand for all the use cases in
        federated and allied environments.




Manral & Hanna                Informational                    [Page 10]

RFC 7018                   Auto-Discovery VPN             September 2013


   16.  The ADVPN solution SHOULD take care of not letting the hub be a
        single point of failure.



        This requirement is driven by demand for all the use cases in
        federated and allied environments.

5.  Security Considerations



   This is a problem statement and requirements document for the
   ADVPN solution and in itself does not introduce any new security
   concerns.  The solution to the problems presented in this document
   may involve dynamic updates to databases defined by RFC 4301,
   such as the Security Policy Database (SPD) or the Peer Authorization
   Database (PAD).

   RFC 4301 is silent about the way these databases are populated, and
   it is implied that these databases are static and preconfigured by a
   human.  Allowing dynamic updates to these databases must be thought
   out carefully because it allows the protocol to alter the security
   policy that the IPsec endpoints implement.

   One obvious attack to watch out for is stealing traffic to a
   particular site.  The IP address for www.example.com is 192.0.2.10.
   If we add an entry to an IPsec endpoint's SPD that says that traffic
   to 192.0.2.10 is protected through peer Gw-Mallory, then this allows
   Gw-Mallory to either pretend to be www.example.com or proxy and read
   all traffic to that site.  Updates to this database require a clear
   trust model.

   Hubs can be a single point of failure that can cause loss of
   connectivity of the entire system; this can be a big security issue.
   Any ADVPN solution design should take care of these concerns.

6.  Acknowledgements



   Many people have contributed to the development of this problem
   statement.  While we cannot thank all contributors, some have played
   an especially prominent role.  Yoav Nir, Yaron Sheffer, Jorge Coronel
   Mendoza, Chris Ulliott, and John Veizades wrote the document upon
   which this specification was based.  Geoffrey Huang, Toby Mao, Suresh
   Melam, Praveen Sathyanarayan, Andreas Steffen, Brian Weis, Lou
   Berger, and Tero Kivinen provided essential input.









Manral & Hanna                Informational                    [Page 11]

RFC 7018                   Auto-Discovery VPN             September 2013


7.  Normative References



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

   [RFC4301]  Kent, S. and K. Seo, "Security Architecture for the
              Internet Protocol", RFC 4301, December 2005.

Authors' Addresses



   Vishwas Manral
   Hewlett-Packard Co.
   3000 Hanover St.
   Palo Alto, CA  94304
   USA

   EMail: vishwas.manral@hp.com


   Steve Hanna
   Juniper Networks, Inc.
   1194 N. Mathilda Ave.
   Sunnyvale, CA  94089
   USA

   EMail: shanna@juniper.net

























Manral & Hanna                Informational                    [Page 12]