RFC 7219






Internet Engineering Task Force (IETF)                        M. Bagnulo
Request for Comments: 7219                            A. Garcia-Martinez
Category: Standards Track                                           UC3M
ISSN: 2070-1721                                                 May 2014


                    SEcure Neighbor Discovery (SEND)
              Source Address Validation Improvement (SAVI)

Abstract



   This memo specifies SEcure Neighbor Discovery (SEND) Source Address
   Validation Improvement (SAVI), a mechanism to provide source address
   validation using the SEND protocol.  The proposed mechanism
   complements ingress filtering techniques to provide a finer
   granularity on the control of IPv6 source addresses.

Status of This Memo



   This is an Internet Standards Track document.

   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).  Further information on
   Internet Standards is available in 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/rfc7219.

Copyright Notice



   Copyright (c) 2014 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.






Bagnulo & Garcia-Martinez    Standards Track                    [Page 1]

RFC 7219                        SEND SAVI                       May 2014


Table of Contents



   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
     1.1.  Requirements Language . . . . . . . . . . . . . . . . . .   4
   2.  Background on SEND SAVI . . . . . . . . . . . . . . . . . . .   4
     2.1.  Address Validation Scope  . . . . . . . . . . . . . . . .   4
     2.2.  Binding Creation for SEND SAVI  . . . . . . . . . . . . .   4
     2.3.  SEND SAVI Protection Perimeter  . . . . . . . . . . . . .   7
     2.4.  Special Cases . . . . . . . . . . . . . . . . . . . . . .   9
   3.  SEND SAVI Specification . . . . . . . . . . . . . . . . . . .  11
     3.1.  SEND SAVI Data Structures . . . . . . . . . . . . . . . .  11
     3.2.  SEND SAVI Device Configuration  . . . . . . . . . . . . .  12
     3.3.  Traffic Processing  . . . . . . . . . . . . . . . . . . .  13
       3.3.1.  Transit Traffic Processing  . . . . . . . . . . . . .  13
       3.3.2.  Local Traffic Processing  . . . . . . . . . . . . . .  13
     3.4.  SEND SAVI Port Configuration Guidelines . . . . . . . . .  27
     3.5.  VLAN Support  . . . . . . . . . . . . . . . . . . . . . .  28
     3.6.  Protocol Constants  . . . . . . . . . . . . . . . . . . .  28
   4.  Protocol Walk-Through . . . . . . . . . . . . . . . . . . . .  29
     4.1.  Change of the Attachment Point of a Host  . . . . . . . .  29
       4.1.1.  Moving to a Port of the Same Switch . . . . . . . . .  29
       4.1.2.  Moving to a Port of a Different Switch  . . . . . . .  30
     4.2.  Attack of a Malicious Host  . . . . . . . . . . . . . . .  31
       4.2.1.  M Attaches to the Same Switch as the Victim's Switch   31
       4.2.2.  M Attaches to a Different Switch to the Victim's
               Switch  . . . . . . . . . . . . . . . . . . . . . . .  32
   5.  Security Considerations . . . . . . . . . . . . . . . . . . .  33
     5.1.  Protection against Replay Attacks . . . . . . . . . . . .  33
     5.2.  Protection against Denial-of-Service Attacks  . . . . . .  34
     5.3.  Considerations on the Deployment Model for Trust Anchors   36
     5.4.  Residual Threats  . . . . . . . . . . . . . . . . . . . .  36
     5.5.  Privacy Considerations  . . . . . . . . . . . . . . . . .  37
   6.  Acknowledgments . . . . . . . . . . . . . . . . . . . . . . .  37
   7.  References  . . . . . . . . . . . . . . . . . . . . . . . . .  37
     7.1.  Normative References  . . . . . . . . . . . . . . . . . .  37
     7.2.  Informative References  . . . . . . . . . . . . . . . . .  38















Bagnulo & Garcia-Martinez    Standards Track                    [Page 2]

RFC 7219                        SEND SAVI                       May 2014


1.  Introduction



   This memo specifies SEND SAVI, a mechanism to provide source address
   validation for IPv6 networks using the SEND protocol [RFC3971].  The
   proposed mechanism complements ingress filtering techniques to
   provide a finer granularity on the control of the source addresses
   used.

   SEND SAVI uses the DAD_NSOL (Duplicate Address Detection Neighbor
   SOLicitation) and the DAD_NADV (DAD Neighbor ADVertisement) messages
   defined in [RFC4862] and the NUD_NSOL (Neighbor Unreachability
   Detection Neighbor SOLicitation) and NUD_NADV (NUD Neighbor
   ADVertisement) messages defined in [RFC4861] to validate the address
   ownership claim of a node.  Using the information contained in these
   messages, host IPv6 addresses are associated to switch ports, so that
   data packets will be validated by checking for consistency in this
   binding, as described in [RFC7039].  In addition, SEND SAVI prevents
   hosts from generating packets containing off-link IPv6 source
   addresses.

   Scalability of a distributed SAVI system comprising multiple SEND
   SAVI devices is preserved by means of a deployment scenario in which
   SEND SAVI devices form a "protection perimeter".  In this deployment
   scenario, the distributed SAVI system only validates the packets when
   they ingress to the protection perimeter, not in every SEND SAVI
   device traversed.

   The SEND SAVI specification, as defined in this document, is limited
   to links and prefixes in which every IPv6 host and every IPv6 router
   uses the SEND protocol [RFC3971] to protect the exchange of Neighbor
   Discovery information.  If the SEND protocol is not used, we can
   deploy other SAVI solutions relying on monitoring different address
   configuration mechanisms to prove address ownership.  For example,
   FCFS (First-Come, First-Served) SAVI [RFC6620] can be used by nodes
   locally configuring IPv6 addresses by means of the Stateless Address
   Autoconfiguration mechanism [RFC4862].

   SEND SAVI is designed to be deployed in SEND networks with as few
   changes to the deployed implementations as possible.  In particular,
   SEND SAVI does not require any changes in the nodes whose source
   address is to be verified.  This is because verification solely
   relies in the usage of already available protocols.  Therefore, SEND
   SAVI neither defines a new protocol nor defines any new message on
   existing protocols, nor does it require that a host or router use an
   existing protocol message in a different way.

   An overview of the general framework about Source Address Validation
   Improvement is presented in [RFC7039].



Bagnulo & Garcia-Martinez    Standards Track                    [Page 3]

RFC 7219                        SEND SAVI                       May 2014


1.1.  Requirements Language



   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.  Background on SEND SAVI



2.1.  Address Validation Scope



   The application scenario of SEND SAVI is limited to the local link.
   This means that the goal of SEND SAVI is to verify that the source
   addresses of the packets generated by the nodes attached to the local
   link have not been spoofed and that only legitimate routers generate
   packets with off-link IPv6 source addresses.

   In a link, there usually are hosts and routers attached.  Hosts
   generate packets with their own addresses as the source address.
   This is called "local traffic".  Routers may send packets containing
   a source address other than their own, since they can forward packets
   generated by other hosts (usually located in a different link).  This
   is the so-called transit traffic.

   SEND SAVI allows the validation of the source address of the local
   traffic, i.e., it allows verification that the source addresses of
   the packets generated by the nodes attached to the local link have
   not been spoofed.  SEND SAVI also provides means to prevent hosts
   from generating packets with source addresses derived from off-link
   prefixes.  However, SEND SAVI does not provide the means to verify if
   a given router is actually authorized to forward packets containing a
   particular off-link source address.  Other techniques, like ingress
   filtering [RFC2827], are recommended to validate transit traffic.

2.2.  Binding Creation for SEND SAVI



   SEND SAVI devices filter packets according to bindings between a
   layer-2 anchor (the binding anchor) and an IPv6 address.  These
   bindings should allow legitimate nodes to use the bounded IPv6
   address as source address and prevent illegitimate nodes from doing
   so.

   Any SAVI solution is not stronger than the binding anchor it uses.
   If the binding anchor is easily spoofable (e.g., a Media Access
   Control (MAC) address), then the resulting solution will be weak.
   The treatment of non-compliant packets needs to be tuned accordingly.
   In particular, if the binding anchor is easily spoofable and the SEND
   SAVI device is configured to drop non-compliant packets, then the
   usage of SEND SAVI may open a new vector of Denial-of-Service (DoS)



Bagnulo & Garcia-Martinez    Standards Track                    [Page 4]

RFC 7219                        SEND SAVI                       May 2014


   attacks, based on spoofed binding anchors.  For that reason,
   implementations of this specification use switch ports as their
   binding anchors.  Other forms of binding anchors are out of the scope
   of this specification, and proper analysis of the implications of
   using them should be performed before their usage.

   SEND [RFC3971] provides tools to assure that a Neighbor Discovery
   (ND) message containing a Cryptographically Generated Address (CGA)
   [RFC3972] option and signed by an RSA option has been generated by
   the legitimate owner of the CGA IPv6 address.

   SEND SAVI uses SEND-validated messages to create bindings between the
   CGA and the port of the SEND SAVI device from which it is reasonable
   to receive packets with the CGA as the source address.  The events
   that trigger the binding creation process in a SEND SAVI device are:

   o  The reception of a DAD_NSOL message, indicating the attempt of a
      node to configure an address.  This may occur when a node
      configures an address for the first time or after being idle for
      some time or when the node has changed the physical attachment
      point to the layer-2 infrastructure.

   o  The reception of any other packet (including data packets) with a
      source address for which no binding exists.  This may occur if
      DAD_NSOL messages were lost, a node has changed the physical
      attachment point to the layer-2 infrastructure without issuing a
      DAD_NSOL message, a SAVI device loses a binding (for example, due
      to a restart), or the link topology changed.

   When the binding creation process is triggered, the SEND SAVI device
   has to assure that the node for which the binding is to be created is
   the legitimate owner of the address.  For the case in which the
   binding creation process is initiated by a DAD_NSOL exchange, the
   SEND SAVI device waits for the reception of a validated DAD_NADV
   message, indicating that the other node has configured the address
   before, or validated DAD_NSOL messages arriving from other locations,
   indicating that another node is trying to configure the same address
   at the same time.  For the case in which packets other than a
   DAD_NSOL initiate the creation of the binding, the SEND SAVI device
   explicitly requires the node sending those packets to prove address
   ownership by issuing a secured NUD_NSOL, which has to be answered
   with a secured NUD_NADV by the probed node.

   SEND SAVI devices issue secured NUD_NSOL messages periodically in
   order to refresh bindings, which have to be answered with a valid
   NUD_NADV message by the node for which the binding exists.





Bagnulo & Garcia-Martinez    Standards Track                    [Page 5]

RFC 7219                        SEND SAVI                       May 2014


   SEND SAVI devices only forward packets with off-link source addresses
   if they are received from a port manually configured to connect to a
   router.

   SEND SAVI needs to be protected against replay attacks, i.e., attacks
   in which a secured SEND message is replayed by another node.  As
   discussed before, the SEND SAVI specification uses SEND messages to
   create a binding between the address contained in the message (that
   must be signed by a node possessing the private key associated to the
   address) and the port through which the message is received.  If an
   attacker manages to obtain such a message from another node, for
   example, because the message was sent to the all-nodes multicast
   address or because the attacker has subscribed to the Solicited Node
   multicast address associated to a remote node, it could replay it
   preserving the original signature.  This may create an illegitimate
   binding in the SEND SAVI device or could be used to abort address
   configuration at the other node.  While SEND provides some means to
   limit the impact of the replay of ND messages, the emphasis for SEND
   anti-replay protection is to limit to a short period of time the
   validity of the ND information transmitted in the message, for
   example, the relationship between an IPv6 address and a layer-2
   address.  Note that the period must be long enough to assure that the
   information sent by the legitimate sender is considered valid despite
   the possible differences in clock synchronization between the sender
   and receiver(s).  For example, with the values recommended by
   [RFC3971] for TIMESTAMP_FUZZ and TIMESTAMP_DRIFT, a node receiving a
   DAD_NSOL message would not discard replays of this message being
   received within a period of approximately 2 seconds (more precisely,
   2/0.99 seconds).  The underlying assumption for SEND security is that
   even if the message is replayed by another node during this period of
   time, the information disseminated by ND is still the same.  However,
   allowing a node to replay a SEND message does have an impact on the
   SEND SAVI operation, regardless of the time elapsed since it was
   generated, since the node can create a new binding in a SEND SAVI
   device for the port to which an illegitimate node attaches.  As can
   be concluded, the protection provided by SEND is not enough in all
   cases for SEND SAVI.

   SEND SAVI increases the protection against the replay attacks
   compared to SEND.  First, each node is required to connect to the
   SEND SAVI topology through a different port to prevent eavesdropping
   before entering the SAVI protection perimeter.  Then, SEND SAVI
   bindings are updated only according to messages whose dissemination
   can be restricted in the SEND SAVI topology without interfering with
   the normal SEND operation.  The messages used by SEND SAVI to create
   bindings are DAD_NSOL messages, for which SEND SAVI limits its
   propagation to the ports through which a previous binding for the
   same IPv6 address existed (see Section 3.3.2), and NUD_NADV messages



Bagnulo & Garcia-Martinez    Standards Track                    [Page 6]

RFC 7219                        SEND SAVI                       May 2014


   in response to a secured NUD_NSOL sent by the SEND SAVI device only
   through the tested port.  Finally, SEND SAVI filtering rules prevent
   nodes from replaying messages generated by the SEND SAVI devices
   themselves.  Section 5.1 discusses in more detail the protection
   provided by SEND SAVI against replay attacks.

2.3.  SEND SAVI Protection Perimeter



   In order to reduce computing and state requirements in SEND SAVI
   devices, SEND SAVI devices can be deployed to form a "protection
   perimeter" [RFC7039].  With this deployment strategy, SEND SAVI
   devices perform source-address validation only when packets enter in
   the protected realm defined through the protection perimeter.  The
   perimeter is defined by appropriate configuration of the roles of
   each port, which can be 'Validating' or 'Trusted':

   o  Validating ports (VPs) are ports in which SEND SAVI filtering and
      binding creation are performed.

   o  Trusted ports (TPs) are ports in which limited processing is
      performed.  Only SEND messages related with certificates, prefix
      information, and DAD operation are processed in order to update
      the state of the SEND SAVI device or the state related with any of
      the Validating ports of the switch.



























Bagnulo & Garcia-Martinez    Standards Track                    [Page 7]

RFC 7219                        SEND SAVI                       May 2014


   Figure 1 shows a typical topology involving trusted and untrusted
   infrastructure.

         +--+   +--+                          +--+   +--+
         |H1|   |H2|                          |H3|   |R1|
         +--+   +--+                          +--+   +--+
           |     |                              |     |
      +------------SEND SAVI PROTECTION PERIMETER-----------+
      |    |     |                              |     |     |
      |  +-1-----2-+                          +-1-----2-+   |
      |  |  SEND-  |                          |  SEND-  |   |
      |  |  SAVI1  |                          |  SAVI2  |   |
      |  +-3--4----+                          +--3--4---+   |
      |    |  |          +--------------+        |  |       |
      |    |  +----------|              |--------+  |       |
      |    |             |   SWITCH-A   |           |       |
      |    |  +----------|              |           |       |
      |    |  |          +--------------+           |       |
      |  +-1--2----+                          +-----1---+   |
      |  |  SEND-  |                          |  SEND-  |   |
      |  |  SAVI3  |                          |  SAVI4  |   |
      |  +-3-----4-+                          +----4----+   |
      |    |     |                                 |        |
      +------------SEND SAVI PROTECTION PERIMETER-----------+
           |     |                                 |
         +--+   +--+                             +--+
         |R2|   |H4|                             |H5|
         +--+   +--+                             +--+



                    Figure 1: SAVI Protection Perimeter

   Trusted ports are used for connections with trusted infrastructures,
   such as routers and other SEND SAVI devices.  Port 2 of SEND-SAVI2
   and port 3 of SEND-SAVI3 are Validating ports because they connect to
   routers.  Port 3 of SEND-SAVI1 and port 1 of SEND-SAVI3 as well as
   port 4 of SEND-SAVI2 and port 1 of SEND-SAVI4 are trusted because
   they connect two SAVI devices.  Finally, port 4 of SEND-SAVI1, port 3
   of SEND-SAVI2, and port 2 of SEND-SAVI3 are trusted because they
   connect to SWITCH-A to which only trusted nodes are connected.

   Validating ports are used for connection with non-trusted
   infrastructures; therefore, hosts connect normally to Validating
   ports.  So, in Figure 1 above, ports 1 and 2 of SEND-SAVI1, port 1 of
   SEND-SAVI2, and port 4 of SEND-SAVI3 are Validating ports because
   they connect to hosts.  Port 4 of SEND-SAVI4 is also a Validating
   port because it is connected to host H5.



Bagnulo & Garcia-Martinez    Standards Track                    [Page 8]

RFC 7219                        SEND SAVI                       May 2014


   For a more detailed discussion on this, see Section 3.4.

2.4.  Special Cases



   Multi-subnet links:  In some cases, a given subnet may have several
      prefixes.  This is supported by SEND SAVI as any port can support
      multiple prefixes.

   Multihomed hosts:  A multihomed host is a host with multiple
      interfaces.  The interaction between SEND SAVI and multihomed
      hosts is as follows.  If the different interfaces of the host are
      assigned different IP addresses and packets sent from each
      interface and always carry the address assigned to that interface
      as the source address, then from the perspective of a SEND SAVI
      device, this is equivalent to two hosts with a single interface,
      each with an IP address.  SEND SAVI supports this without
      additional considerations.  If the different interfaces share the
      same IP address or if the interfaces have different addresses but
      the host sends packets using the address of one of the interfaces
      through any of the interfaces, then SEND SAVI does not directly
      support it.  It would require either connecting at least one
      interface of the multihomed host to a Trusted port or manually
      configuring the SEND SAVI bindings to allow binding the address of
      the multihomed host to multiple anchors simultaneously.

   Virtual switches:  A hypervisor or a host operating system may
      perform bridging functions between virtual hosts running on the
      same machine.  The hypervisor or host OS may in turn connect to a
      SEND SAVI system.  This scenario is depicted in Figure 2, with two
      virtual machines, VM1 and VM2, connected through a virtual switch,
      VS1, to SEND SAVI device SEND-SAVI1.  The attachment points of VS1
      to VM1 and VM2 are configured as Validating.



















Bagnulo & Garcia-Martinez    Standards Track                    [Page 9]

RFC 7219                        SEND SAVI                       May 2014


       Host1
       +----------------+
       | +---+   +---+  |
       | |VM1|   |VM2|  |
       | +---+   +---+  |
       |   |     |      |
       | +-1-----2--+   |
       | |   VS1    |   |
       | +--3-------+   |
       |    |           |
       +----|-----------+
            |
            |
         +--1-----2--+
         |   SEND-   |
         |   SAVI1   |
         +--3---4----+
            |   |

       Figure 2: Virtual Switches Connected to the SEND SAVI Device

   In order to provide proper security against replay attacks,
   performing SEND SAVI filtering as close to untrusted hosts as
   possible (see Sections 3.4 and 5.1) is recommended.  In this
   scenario, this objective can be achieved by enabling SEND SAVI
   validation in VS1.  Ideally, VS1 could be integrated into the SEND
   SAVI protection perimeter if the hypervisor or host OS at Host1 can
   be trusted (even though VM1 and VM2 could not be trusted).  To do so,
   both the attachment to SEND-SAVI1 at VS1, and port 1 at SEND-SAVI1,
   are configured as Trusted.

   If the administrator of the network does not trust VS1, port 1 of
   SEND-SAVI1 is configured as Validating, so that every address being
   used at Host1 is validated at SEND-SAVI1 by SEND SAVI.  The
   attachment point to the physical network at VS1 should be configured
   as Trusted if the host administrator knows that it is connected to a
   SEND SAVI device; in this case, VS1 relies on the infrastructure
   comprised by the physical SEND SAVI devices but not vice versa.
   Packets egressing from VM1 are validated twice: first at VS1 and then
   at SEND-SAVI1.  Packets going in the reverse direction (from an
   external host to VM1) are validated once: when they first reach a
   SEND SAVI device.  If the administrator of VS1 does not trust the
   physical switch to which it attaches, it can configure the attachment
   to SEND-SAVI1 as Validating.  In Figure 2 above, this means that a
   packet going from another host to VM1 would be validated twice: once
   when entering the SEND SAVI perimeter formed by the physical devices
   and again when entering at VS1.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 10]

RFC 7219                        SEND SAVI                       May 2014


   Untrusted routers: One can envision scenarios where routers are
   dynamically attached to a SEND SAVI network.  A typical example would
   be a mobile phone connecting to a SEND SAVI switch where the mobile
   phone is acting as a router for other personal devices that are
   accessing the network through it.  Regarding the validation of the
   source address performed in a SEND SAVI device, such an untrusted
   router does not seem to directly fall in the category of trusted
   infrastructure (if this was the case, it is likely that all devices
   would be trusted); hence, it cannot be connected to a Trusted port,
   and if it is connected to a Validating port, the SEND SAVI switch
   would discard all the packets containing an off-link source address
   coming from that device.  Although the SEND SAVI device to which this
   router attaches could be configured to permit the transit of packets
   with source addresses belonging to the set of prefixes reachable
   through the untrusted router, such a mechanism is out of the scope of
   this document.  As a result, the default mechanism described in this
   specification cannot be applied in such a scenario.

3.  SEND SAVI Specification



3.1.  SEND SAVI Data Structures



   The following three data structures are defined for SEND SAVI
   operations.

   SEND SAVI Database: The SEND SAVI function relies on state
   information binding the source IPv6 address used in data packets to
   the port through which the legitimate node connects.  Such
   information is stored in the SEND SAVI Database.  The SEND SAVI
   Database is populated with the contents of validated SEND messages.
   Each entry contains the following information:

   o  IPv6 source address

   o  Binding anchor: the port through which the packet was received

   o  Lifetime

   o  Status: TENTATIVE_DAD, TENTATIVE_NUD, VALID, TESTING_VP,
      TESTING_VP'

   o  Alternative binding anchor: the port from which a DAD_NSOL message
      or any data packet has been received while a different port was
      stored in the binding anchor for the address.

   o  Creation time: the value of the local clock when the entry was
      first created




Bagnulo & Garcia-Martinez    Standards Track                   [Page 11]

RFC 7219                        SEND SAVI                       May 2014


   SEND SAVI Prefix List: SEND SAVI devices need to know which ones are
   the link prefixes in order to identify local and off-link traffic.  A
   SEND SAVI device MUST support discovering this information from the
   Prefix Information option [RFC4861] with the L bit set of Router
   Advertisement (RADV) messages coming from Trusted ports, as described
   in Section 3.3.2.  The list of prefixes MAY also be configured
   manually.  This information is not specific to a given port.  The
   SEND SAVI Prefix List contains one entry per prefix in use, as
   follows:

   o  Prefix: the prefix included in a Prefix Information option.

   o  Prefix lifetime: time in seconds that the prefix is valid.
      Initially set to the Valid Lifetime value of the Prefix
      Information option of a valid RADV message or set to a value of
      all 1 bits (0xffffffff), which represents infinity, if configured
      manually.

   When the SEND SAVI device boots, it MUST send a Router Solicitation
   (RSOL) message, which does not need to be secured if the unspecified
   address is used (see [RFC3971], Sections 5.1.1 and 5.2.1).  The SAVI
   device SHOULD issue a RSOL message in case the prefix entry is about
   to expire.

3.2.  SEND SAVI Device Configuration



   In order to perform the SEND SAVI operation, some basic parameters of
   the SEND SAVI device have to be configured.  Since a SEND SAVI device
   operates as a SEND node to generate NUD_NSOL, RSOL, or Certification
   Path Solicitation (CPS) messages:

   o  The SEND SAVI device MUST be configured with a valid CGA address.
      When the SEND SAVI device configures this address, it MUST behave
      as a regular SEND node, i.e., using secured NSOL messages to
      perform DAD, etc., in addition to fulfilling the requirements
      stated for regular IPv6 nodes [RFC6434].

   o  The SEND SAVI device MAY be configured with at least one trust
      anchor if it is configured to validate RADV messages (see
      Section 3.3.2).  In this case, the SEND SAVI device MAY be
      configured with certification paths.  The alternative is obtaining
      them by means of issuing Certification Path Solicitation messages,
      as detailed in the SEND specification [RFC3971].

   In addition, the port role for each port of the SEND SAVI device MUST
   be configured.  The guidelines for this configuration are specified
   in Section 3.4.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 12]

RFC 7219                        SEND SAVI                       May 2014


3.3.  Traffic Processing



   In this section, we describe how packets are processed by a SEND SAVI
   device.  Behavior varies depending on if the packet belongs to local
   or transit traffic.  This is determined by checking if the prefix of
   the source address is included in the SEND SAVI Prefix List or in the
   unspecified address (local traffic) or not included in the SEND SAVI
   Prefix List (transit traffic).

3.3.1.  Transit Traffic Processing



   Transit traffic processing occurs as follows:

   o  If the SEND SAVI device receives a transit traffic packet through
      a Trusted port, it forwards it without any SAVI processing.

   o  If the SEND SAVI device receives a transit traffic packet through
      a Validating port, it discards the packet.

3.3.2.  Local Traffic Processing



   If the verification of the source address of a packet shows that it
   belongs to local traffic, this packet is processed using the state
   machine described in this section.

   For the rest of the section, the following assumptions hold:

   o  When it is stated that a secured NUD_NSOL message is issued by a
      SEND SAVI device through a port P, it means that the SEND SAVI
      device generates a NUD_NSOL message, according to the Neighbor
      Unreachability Detection procedure described in [RFC4861],
      addressed to the IPv6 target address, which is the source address
      of the packet triggering the procedure.  This message is secured
      by SEND as defined in [RFC3971].  The source address used for
      issuing the NUD_NSOL message is the source address of the SEND
      SAVI device.  The message is sent only through port P.

   o  When it is stated that a validated NUD_NADV message is received by
      a SEND SAVI device, it means that a SEND secured NUD_NADV message
      has been received by the same port P through which the
      corresponding NUD_NSOL message was issued, and the NUD_NADV
      message has been validated according to [RFC3971] to prove
      ownership for the IPv6 address under consideration and to prove
      that it is a response for the previous NUD_NSOL message issued by
      the SEND SAVI device (containing the same nonce value as the
      NUD_NSOL message to which it answers).





Bagnulo & Garcia-Martinez    Standards Track                   [Page 13]

RFC 7219                        SEND SAVI                       May 2014


   We use VP to refer to a Validating port and TP to refer to a Trusted
   port.

   The state machine is defined for a binding of a given source IPv6
   address in a given SEND SAVI device.  In the transitions considered,
   packets described as inputs refer to the IPaddr IPv6 address
   associated to the state machine.

   The possible states for a given IPaddr are NO_BIND, TENTATIVE_DAD,
   TENTATIVE_NUD, VALID, TESTING_VP, and TESTING_VP'.  The NO_BIND state
   represents that no binding exists for IPaddr; this is the state for
   all addresses unless a binding is explicitly created.

   The states can be classified into 'forwarding' states, i.e., states
   in which packets received from the port associated to the IPv6
   address are forwarded, and 'non-forwarding' states, i.e., states in
   which packets different to the ones used for signaling are not
   forwarded.  VALID, TENTATIVE_DAD, TESTING_VP, and TESTING_VP' are
   forwarding states, and NO_BIND and TENTATIVE_NUD are non-forwarding
   states.

   The SEND SAVI device MUST join the Solicited Node Multicast group for
   all the addresses whose state is other than NO_BIND.  This is needed
   to make sure that the SEND SAVI device receives DAD_NSOL messages
   issued for those addresses.  Note that it may not be enough to relay
   on the Multicast Listener Discovery (MLD) messages being sent by the
   node attached to a Validating port for which a binding for the
   corresponding address exists, since the node may move and packets
   sent to that particular Solicited Node Multicast group may no longer
   be forwarded to the SEND SAVI device.

   In order to determine which traffic is on-link and off-link, the SEND
   SAVI device MUST support discovery of this information from the
   Prefix Information option with the L bit set of RADV messages.  In
   this case, at least one router SHOULD be configured to advertise RADV
   messages containing a Prefix Information option with the prefixes
   that the untrusted nodes can use as source addresses, and the bit L
   set.  An alternative to this is to manually configure the SEND SAVI
   Prefix List or restrict the use of link-local addresses.

   SEND SAVI devices MUST discard RADV messages received from Validating
   ports.  RADV messages are only accepted and processed when received
   through Trusted ports.

   SEND SAVI devices SHOULD NOT validate RADV messages to update the
   SEND SAVI Prefix List and forward them to other nodes.  These
   messages can only be received from Trusted ports, and we assume that
   routers are trusted.  Validating RADV messages would be required in



Bagnulo & Garcia-Martinez    Standards Track                   [Page 14]

RFC 7219                        SEND SAVI                       May 2014


   any SEND SAVI device the node is traversing.  Besides, hosts will
   validate this message before using the information it contains.

   In case SEND SAVI devices are configured to validate RADV messages,
   SEND SAVI devices SHOULD support the processing of validated
   Certification Path Advertisement (CPA) messages, sent in reply to CPS
   messages, to acquire certificates used to validate router messages;
   alternatively, it SHOULD be configured with a certification path.

   The state machine defined for the SEND SAVI operation adheres to the
   following design guidelines:

   o  The only events that trigger state changes from forwarding to non-
      forwarding states, and vice versa, are the reception of DAD_NSOL,
      DAD_NADV, and NUD_NADV or the expiration of a timer.  The other
      possible input to consider is 'any other packet', which could
      generate changes to states belonging to the same forwarding or
      non-forwarding class as the original state.  In other words, when
      'any other packet' is received, the state cannot move from
      forwarding to non-forwarding, and vice versa.  The reduced set of
      messages being able to trigger a change simplifies the processing
      at SEND SAVI devices.

   o  DAD_NADV and NUD_NADV are only processed when they are a response
      to a DAD_NSOL or a NUD_NSOL message.

   o  SEND SAVI devices MUST only use ND messages received through
      Validating ports if they are valid; otherwise, they discard them.
      SEND SAVI devices SHOULD assume that such messages received from
      Trusted ports have been validated by other SEND SAVI devices, or
      come from a trusted device such a router, so they SHOULD NOT
      attempt to validate them in order to reduce the processing load at
      the SEND SAVI device.

   o  The only messages the SEND SAVI device is required to generate
      specifically per each source IP address are MLD and NUD_NSOL
      messages.  This also keeps the state machine simple.

   o  Well-behaved nodes are expected to initiate communication by
      sending secured DAD_NSOL messages.  The SEND SAVI state machine is
      tailored to efficiently process these events.  The reception of
      other packet types without receiving previously validated DAD_NSOL
      messages is assumed to be a consequence of bad-behaving nodes or
      infrequent events (such as packet loss, a change in the topology
      connecting the switches, etc.).  While a binding will ultimately
      be created for nodes affected by such events, simplicity of the
      state machine is prioritized over any possible optimization for
      these cases.



Bagnulo & Garcia-Martinez    Standards Track                   [Page 15]

RFC 7219                        SEND SAVI                       May 2014


   o  If a node has a configured address, and it can prove that it owns
      this address, the binding is preserved regardless of any
      indication that a binding for the same source address could be
      configured in other SEND SAVI devices.  Bindings for the same
      source address in two or more SEND SAVI devices may occur due to
      several reasons, for example, when a host moves (the two bindings
      exist just for a short period of time) or when many nodes generate
      the same address and the DAD procedure has failed.  In these
      infrequent cases, SEND SAVI preserves connectivity for the
      resulting bindings.

   Next, we describe how different inputs are processed, depending on
   the state of the binding of the IP address 'IPaddr'.  Note that every
   ND message is assumed to be validated according to the SEND
   specification.

   To facilitate the reader's understanding of the most relevant
   transitions of the SEND SAVI state machine, a simplified version,
   which does not contain every possible transition, is depicted in
   Figure 3:































Bagnulo & Garcia-Martinez    Standards Track                   [Page 16]

RFC 7219                        SEND SAVI                       May 2014


                          +-------------+
                          |             |
                          | TESTING_VP' |
                          |             |
                          +-------------+
             Timeout/VP=VP'  |    ^
                             |    |
             VP_NUD_NADV/-   |    |  VP'_DAD_NSOL/
                             |    |    VP_NUD_NSOL
                             |    |
                             v    |
         VP_DAD_NSOL/-     +--------+
            +------------- |        |
            |              | VALID  |< -------------------+
            |   +-------- >|        |                     |
            |   |          +--------+                     |
            |   |            ^   |                        |
            |   |    VP_NUD_ |   | Timeout,               |
            |   |     NADV/- |   | TP_DAD_NSOL/VP_NUD_NSOL|
            |   |            |   v                        |
            |   |         +------------+                  |
            |   |         |            |                  |
            |   |         | TESTING_VP |                  |
            |   |         |            |                  |
            |   |         +------------+                  |
            |   |              |                          |
            |   |              | Timeout/-                |
            |   | VP*,         |                          |
            |   | Timeout/-    |            VP_NUD_NADV/- |
            v   |              |                          |
         +---------------+     |           +---------------+
         |               |     |           |               |
         | TENTATIVE_DAD |     |           | TENTATIVE_NUD |
         |               |     |           |               |
         +---------------+     |           +---------------+
            ^  |               |             |         ^
            |  |               |   Timeout/- |         |
            |  | TP_DAD_NSOL,  |             |         |
            |  | TP_DAD_NADV/- |             |         |
            |  |               v             |         |
            |  |           +---------+       |         |
            |  +--------- >|         |< -----+         |
            |              | NO_BIND |                 |
            +--------------|         |-----------------+
            VP_DAD_NSOL/-  +---------+    VP*/VP_NUD_NSOL

               Figure 3: Simplified SEND SAVI State Machine




Bagnulo & Garcia-Martinez    Standards Track                   [Page 17]

RFC 7219                        SEND SAVI                       May 2014


   Each state transition is characterized by any of the events that may
   trigger the change and the message(s) generated as a result of this
   change.  The meaning of some terms are referred next:

   o  VP_DAD_NSOL as a triggering event means that a validated DAD_NSOL
      message has been received from the current BINDING_ANCHOR port VP.

   o  VP* means any packet (data packet) received from the current
      BINDING_ANCHOR port VP.

   o  TP_DAD_NSOL as a triggering event means that a DAD_NSOL message
      was received from a Trusted port.

   o  - means that no message is sent.  VP=VP' means that the
      BINDING_ANCHOR is set to VP'.

   The notation

      Timeout, TP_DAD_NSOL/VP_NUD_NSOL

   means that the transition is triggered by either a timeout expiration
   or the reception of a DAD_NSOL message from a Trusted port, and in
   addition to the transition, a NUD_NSOL message is sent through port
   VP.

   For the rest of the description, we assume the following:

   o  When a validated message is required (i.e., a 'validated
      DAD_NSOL'), messages are check for validity in the considered
      switch according to [RFC3971], and messages not fulfilling these
      conditions are discarded.

   o  When any SEND message is received from a validated port, the SEND
      SAVI SHOULD assume that the message has been validated by the SEND
      SAVI device through which the message accessed the SEND SAVI
      protection perimeter (unless the SEND SAVI perimeter has been
      breached), or the device generating it is trusted.  In this case,
      the SAVI device does not perform any further validation.
      Performing validation for SEND messages received through a Trusted
      port may affect performance negatively.











Bagnulo & Garcia-Martinez    Standards Track                   [Page 18]

RFC 7219                        SEND SAVI                       May 2014


   NO_BIND

   When the node is in this state, there are no unresolved NUD_NSOL
   messages generated by SEND SAVI or DAD_NSOL propagated to any
   Validating port, so the only relevant inputs are DAD_NSOL messages
   coming either from a Validating port (VP) or Trusted port (TP), or
   any packet other than DAD_NSOL coming from a VP or TP.  There are no
   timers configured for this state.

   Messages received from a Validating port:

   o  If a validated DAD_NSOL message is received from a Validating port
      VP, the SEND SAVI device forwards this message to all appropriate
      Trusted ports (the subset of Trusted ports that belong to the
      forwarding layer-2 topology, with the restrictions imposed by the
      MLD snooping mechanism, if applied).  DAD_NSOL messages are not
      sent through any of the ports configured as Validating ports.  The
      SEND SAVI device sets the LIFETIME to TENT_LT, stores all the
      information required for future validation of the corresponding
      DAD_NADV message (such as the nonce of the message), creates a new
      entry in the SEND SAVI Database for IPaddr, sets BINDING_ANCHOR to
      VP, and changes the state to TENTATIVE_DAD.  Creation time is set
      to the current value of the local clock.

      Note that in this case, it is not possible to check address
      ownership by sending a NUD_NSOL because while the node is waiting
      for a possible DAD_NADV, its address is in tentative state and the
      node cannot respond to NSOL messages [RFC4862].

   o  If any packet other than a DAD_NSOL is received through a
      Validating port VP, the SEND SAVI device issues a secured NUD_NSOL
      through port VP.  The SEND SAVI device sets the LIFETIME to
      TENT_LT.  The SEND SAVI device creates a new entry in the SEND
      SAVI Database for IPaddr, sets BINDING_ANCHOR to VP, and the state
      is changed to TENTATIVE_NUD.  Creation time is set to the current
      value of the local clock.  The SAVI device MAY discard the packet
      while the NUD procedure is being executed or MAY store it in order
      to send it if the next transitions are (strictly) TENTATIVE_NUD
      and then VALID.

   Messages received from a Trusted port:

   o  If a DAD_NSOL message containing IPaddr as the target address is
      received through a Trusted port, it MUST NOT be forwarded through
      any of the Validating ports: it is sent through the proper Trusted
      ports.  The state is not changed.





Bagnulo & Garcia-Martinez    Standards Track                   [Page 19]

RFC 7219                        SEND SAVI                       May 2014


   o  Any packet other than a DAD_NSOL received from a Trusted port is
      forwarded to its destination.  This packet is assumed to come from
      a SEND SAVI device that has securely validated the binding,
      according to the SEND SAVI rules (unless the SEND SAVI perimeter
      has been breached).  The state is not changed.

   TENTATIVE_DAD

   To arrive at this state, the SEND SAVI device has received a
   validated DAD_NSOL coming from the BINDING_ANCHOR port, and it has
   forwarded it to the appropriate TPs.  The relevant events occurring
   in this state are the reception of a DAD_NADV message from a TP, a
   DAD_NSOL message from the BINDING_ANCHOR port, other Validating port
   or TP, a data packet from the BINDING_ANCHOR port, and the expiration
   of the LIFETIME timer initiated when the DAD_NSOL was received at the
   BINDING_ANCHOR port.

   Messages received from a Trusted port:

   o  The reception of a valid DAD_NADV message from a Trusted port
      indicates that the binding cannot be configured for the
      BINDING_ANCHOR port.  The state is changed to NO_BIND, and the
      LIFETIME is cleared.

   o  The reception of a valid DAD_NSOL from a Trusted port indicates
      that a node connected to another SEND SAVI device may be trying to
      configure the same address at the same time.  The DAD_NSOL message
      is forwarded to the BINDING_ANCHOR port, so that the node at this
      port will not configure the address, as stated in [RFC4862].  The
      DAD_NSOL message is also forwarded to all appropriate Trusted
      ports.  Then, the LIFETIME is cleared, and the state is changed to
      NO_BIND.

   o  Any packet other than a validated DAD_NSOL or DAD_NADV received
      from a Trusted port is forwarded to its destination.  This packet
      is assumed to come from a SEND SAVI device that has securely
      validated the binding, according to the SEND SAVI rules (unless
      the SEND SAVI perimeter has been breached).  The state is not
      changed.












Bagnulo & Garcia-Martinez    Standards Track                   [Page 20]

RFC 7219                        SEND SAVI                       May 2014


   Messages received from a Validating port different from the
   BINDING_ANCHOR:

   o  A validated DAD_NSOL is received from a Validating port VP'
      different from the BINDING_ANCHOR port.  The reception of a valid
      DAD_NSOL from port VP' indicates that a node connected to VP' may
      be trying to configure the same address at the same time.  The
      DAD_NSOL message is forwarded to the BINDING_ANCHOR port, so that
      the node at this port will not configure the address, as stated in
      [RFC4862].  The DAD_NSOL message is also forwarded to all
      appropriate Trusted ports.  Then, the BINDING_ANCHOR is set to VP'
      (through which the DAD_NSOL message was received), the LIFETIME is
      set to TENT_LT, and the state remains in TENTATIVE_DAD.

   o  Any packet other than a validated DAD_NSOL received from a
      Validating port VP' different from the BINDING_ANCHOR port is
      discarded.  The state is not changed.

   Messages received from the BINDING_ANCHOR port:

   o  If a validated DAD_NSOL is received from the BINDING_ANCHOR port,
      the LIFETIME is set to TENT_LT, and the state remains in
      TENTATIVE_DAD.

   o  If any packet other than a DAD_NSOL is received from the
      BINDING_ANCHOR port, it is assumed that the node has configured
      its address, although it has done it in less time than expected by
      the SEND SAVI device (less than TENT_LT).  Since the node proved
      address ownership by means of the validated DAD_NSOL message, the
      LIFETIME is set to DEFAULT_LT, and the state is changed to VALID.

   LIFETIME expires:

   o  If LIFETIME expires, it is assumed that no other node has
      configured this address.  Therefore, the Validating port VP
      (currently stored in the BINDING_ANCHOR) could be bound to this
      IPv6 address.  The LIFETIME is set to DEFAULT_LT, and the state is
      changed to VALID.

   VALID

   To arrive at this state, the SEND SAVI device has successfully
   validated address ownership and has created a binding for IPaddr.
   Relevant transitions for this state are triggered by the reception of
   DAD_NSOL from the BINDING_ANCHOR port, other Validating port or a TP,
   and any packet other than DAD_NSOL from a Validating port other than





Bagnulo & Garcia-Martinez    Standards Track                   [Page 21]

RFC 7219                        SEND SAVI                       May 2014


   the BINDING_ANCHOR or a TP.  The expiration of LIFETIME is also
   relevant to trigger a check for address ownership for the node at the
   BINDING_ANCHOR port.

   Messages received from the BINDING_ANCHOR port:

   o  If a validated DAD_NSOL with IPaddr as a source address is
      received through the BINDING_ANCHOR port, it is forwarded to the
      appropriate Trusted ports.  The LIFETIME is set to TENT_LT, and
      the state is changed to TENTATIVE_DAD.

   o  Any packet other than a DAD_NSOL containing IPaddr as a source
      address arriving from the BINDING_ANCHOR port is forwarded
      appropriately.  The state is not changed.

   Messages received from a Trusted port:

   o  If a DAD_NSOL with IPaddr as a source address is received through
      a Trusted port, the message is forwarded to VP.  The LIFETIME is
      set to TENT_LT, a secured NUD_NSOL message is sent to IPaddr
      through VP, and the state is changed to TESTING_VP.

   o  If any packet other than a DAD_NSOL with IPaddr as a source
      address is received through a Trusted port, the packet is
      forwarded to VP and to other appropriate Trusted ports.  A secured
      NUD_NSOL is sent to the BINDING_ANCHOR port, the LIFETIME is set
      to TENT_LT, and the state is changed to TESTING_VP.

   Messages received from a Validating port different from the
   BINDING_ANCHOR:

   o  If a validated DAD_NSOL packet with IPaddr as a source address is
      received through a Validating port VP' (a VP' different from the
      current BINDING ANCHOR), the message is forwarded to the
      BINDING_ANCHOR port.  In addition, a secured NUD_NSOL is sent to
      the BINDING_ANCHOR port, the ALTERNATIVE BINDING ANCHOR is set to
      port VP' (for future use if the node at VP' is finally selected),
      the LIFETIME is set to TENT_LT, and the state is changed to
      TESTING_VP'.

   o  If any packet other than a DAD_NSOL with IPaddr as a source
      address is received from a Validating port VP', different from the
      current BINDING_ANCHOR for this binding, VP, the packet is
      discarded.  The SEND SAVI device MAY issue a secured NUD_NSOL
      through the BINDING_ANCHOR port, store VP' in the ALTERNATIVE
      BINDING ANCHOR for possible future use, set the LIFETIME to
      TENT_LT, and change the state to TESTING_VP'.  An alternative to
      this behavior is that the SEND SAVI device MAY not do anything (in



Bagnulo & Garcia-Martinez    Standards Track                   [Page 22]

RFC 7219                        SEND SAVI                       May 2014


      this case, the state would eventually change after a maximum
      DEFAULT_LT time; if the node at VP does not respond to a NUD_NSOL
      at TESTING_VP, the state is moved to NO_BIND).  Then, a packet
      arriving from VP' would trigger a process that may end up with
      binding for the node connecting to VP'.

   LIFETIME expires:

   o  If LIFETIME expires, a secured NUD_NSOL message is sent through
      the BINDING_ANCHOR port to IPaddr, the LIFETIME is set to TENT_LT,
      and the state is changed to TESTING_VP.  In the TESTING_VP state,
      packets are still being forwarded until the timer expires without
      receiving a NUD_NADV.

   TESTING_VP

   When the SEND SAVI device enters the TESTING_VP state, the current
   Validating port is under check through a secured NUD_NSOL message
   generated by the SEND SAVI device.  While testing, packets from the
   current Validating port are forwarded.  Packets coming from Trusted
   ports are also forwarded.  The relevant events for this state are the
   reception of a NUD_NADV message from VP; the reception of a DAD_NSOL
   message from VP, VP', or TP; the reception of any packet other than
   the previous cases from VP, VP', or TP; and the expiration of the
   timer associated to the reception of NUD_NADV.

   Messages received from the BINDING_ANCHOR port:

   o  If a validated NUD_NADV is received from VP, the LIFETIME is
      changed to DEFAULT_LT, and the state is changed to VALID.  The
      message is not forwarded to any other port.

   o  If a validated DAD_NSOL message is received from VP, it is
      forwarded to the appropriate Trusted ports, the LIFETIME is set to
      DEFAULT_LT, and the state is changed to TENTATIVE_DAD.

   o  Any packet other than DAD_NSOL or NUD_NADV containing IPaddr as a
      source address arriving from the BINDING_ANCHOR port is forwarded.
      Neither the LIFETIME nor the state are changed.

   Messages received from a Trusted port:

   o  If a DAD_NSOL packet is received from a Trusted port, the message
      is forwarded to VP and the appropriate Trusted ports.  Neither the
      LIFETIME nor the state are changed.  The node at the
      BINDING_ANCHOR port is under check; if it still is at this port,
      it should answer with a NUD_NADV and also with a DAD_NADV.  If it
      is not there, neither the NUD_NADV nor the DAD_NADV will be



Bagnulo & Garcia-Martinez    Standards Track                   [Page 23]

RFC 7219                        SEND SAVI                       May 2014


      received, the timer will expire, and the local state will move to
      NO_BIND.

   o  If a packet other than a DAD_NSOL arrives from a Trusted port, the
      packet is forwarded.  Neither the LIFETIME nor the state are
      changed.

   Messages received from a Validating port different from the
   BINDING_ANCHOR:

   o  If a valid DAD_NSOL is received from a Validating port VP' other
      than the current BINDING_ANCHOR port, the message is forwarded to
      the BINDING_ANCHOR port and to the appropriate Trusted ports.  In
      addition, a secured NUD_NSOL is sent to the BINDING_ANCHOR port,
      the ALTERNATIVE BINDING ANCHOR is set to VP' (for future use if
      the node at VP' is finally selected), the LIFETIME is set to
      TENT_LT, and the state is changed to TESTING_VP'.

   o  Any other packet received from a Validating port VP' other than
      the BINDING_ANCHOR port is discarded.  This may occur because the
      node has moved but has not issued a DAD_NSOL or the DAD_NSOL
      message has been lost.  The state will eventually move to NO_BIND,
      and then the packets sent from VP' will trigger the creation of
      the binding for VP'.

   LIFETIME expires:

   o  If the LIFETIME expires, the LIFETIME is cleared and the state is
      changed to NO_BIND.

   TESTING_VP'

   To arrive at this state, the SEND SAVI device has received an
   indication that a node at VP' different from the BINDING_ANCHOR port
   wants to send data with IPaddr as a source address and has occurred
   while a binding existed for VP.  The port VP' that triggered the
   change of the state to TESTING_VP' was stored at the
   ALTERNATIVE_BINDING_ANCHOR, so that it can be retrieved if the node
   at VP' is determined as the legitimate owner of IPaddr.  The SEND
   SAVI device has issued a NUD_NSOL to IPaddr through the
   BINDING_ANCHOR port.  The relevant events that may occur in this case
   are the reception of a NUD_NADV from port VP (the BINDING_ANCHOR
   port); the reception of a DAD_NSOL from VP, VP', TP, and VP" (VP"
   different from VP and VP'); the reception of any other packet from
   VP, VP', TP, or VP"; and the expiration of the timer.






Bagnulo & Garcia-Martinez    Standards Track                   [Page 24]

RFC 7219                        SEND SAVI                       May 2014


   Messages received from the BINDING_ANCHOR port:

   o  A validated NUD_NADV is received from the BINDING_ANCHOR port.
      The reception of a valid NUD_NADV indicates that the node at VP is
      defending its address.  The BINDING_ANCHOR in use is kept, the
      LIFETIME is set to DEFAULT_LT, and the state is changed to VALID.

   o  If a valid DAD_NSOL is received from the BINDING_ANCHOR port, it
      is forwarded to VP' (the port stored in the
      ALTERNATIVE_BINDING_ANCHOR).  The BINDING_ANCHOR in use is kept,
      the LIFETIME is set to TENT_LT, and the state is changed to
      TENTATIVE_DAD.  When the DAD_NSOL message is received by the node
      at VP', the address will not be configured.

   o  Any packet other than a validated DAD_NSOL, or a validated
      NUD_NADV coming from the BINDING_ANCHOR port, is forwarded, and
      the state is not changed.

   Messages received from the ALTERNATIVE_BINDING_ANCHOR Validating
   port:

   o  If a valid DAD_NSOL is received from the port stored in the
      ALTERNATIVE_BINDING_ANCHOR, it is forwarded to the BINDING_ANCHOR
      port.  The BINDING_ANCHOR and the ALTERNATIVE BINDING ANCHOR are
      kept, the LIFETIME is set to DEFAULT_LT, and the state is not
      changed.

   o  Any packet other than a validated DAD_NSOL coming from the
      ALTERNATIVE_BINDING_ANCHOR port is discarded, and the state is not
      changed.

   Messages received from a Validating port different from the
   BINDING_ANCHOR and the ALTERNATIVE_BINDING_ANCHOR ports:

   o  If a validated DAD_NSOL is received from port VP", different from
      BINDING_ANCHOR and the ALTERNATIVE_BINDING_ANCHOR ports, it is
      forwarded to the BINDING_ANCHOR and the ALTERNATIVE_BINDING_ANCHOR
      ports.  The node at the ALTERNATIVE BINDING ANCHOR port is
      expected to unconfigure its address if the message triggering the
      transition to this state was a DAD_NSOL message received from the
      ALTERNATIVE_BINDING_ANCHOR port (and not any other packet).  The
      state remains in TESTING_VP', although VP" is stored in the
      ALTERNATIVE_BINDING_ANCHOR for future use if the node at VP" is
      finally selected.  The LIFETIME is not changed.

   o  Any packet other than a validated DAD_NSOL received from port VP"
      is discarded and does not affect the state.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 25]

RFC 7219                        SEND SAVI                       May 2014


   Messages received from a Trusted port:

   o  If a DAD_NSOL is received from a Trusted port, the message is
      forwarded to the BINDING_ANCHOR, ALTERNATIVE_BINDING_ANCHOR ports,
      and other appropriate Trusted ports.  The LIFETIME is left
      unchanged, and the state is changed to TESTING_VP.  The node at
      the ALTERNATIVE_BINDING_ANCHOR port is expected to unconfigure its
      address if the packet triggering the transition to this state was
      a DAD_NSOL message received from the ALTERNATIVE_BINDING_ANCHOR
      port.

   o  Any packet other than a DAD_NSOL coming from a Trusted port is
      forwarded appropriately, but the state is not changed.

   LIFETIME expires:

   o  If LIFETIME expires, it is assumed that the node for which the
      binding existed is no longer connected through the BINDING_ANCHOR
      port.  Therefore, the BINDING_ANCHOR is set to the
      ALTERNATIVE_BINDING_ANCHOR port value.  The LIFETIME is set to
      DEFAULT_LT, and the state is changed to VALID.

   TENTATIVE_NUD

   To arrive at this state, a data packet has been received through the
   BINDING_ANCHOR port without any existing binding in the SEND SAVI
   device.  The SEND SAVI device has sent a NUD_NSOL message to the
   BINDING_ANCHOR port.  The relevant events for this case are the
   reception of a NUD_NADV from the BINDING_ANCHOR port; the reception
   of a DAD_NSOL from the BINDING_ANCHOR port, other VP different from
   the BINDING_ANCHOR port, or a TP; and the reception of any packet
   other than a DAD_NSOL and a NUD_NADV from the BINDING_ANCHOR port and
   a DAD_NSOL for other VP different from the BINDING_ANCHOR port, or
   TP.  In addition, the LIFETIME may expire.

   Messages received from the BINDING_ANCHOR port:

   o  If a validated NUD_NADV message is received through the
      BINDING_ANCHOR port, the LIFETIME is set to TENT_LT, and the state
      is changed to VALID.  The message is not forwarded to any port.

   o  If a validated DAD_NSOL message is received through the
      BINDING_ANCHOR port, it is forwarded to the appropriate Trusted
      ports, the LIFETIME is set to TENT_LT, and the state is changed to
      TENTATIVE_DAD.

   o  Any packet other than NUD_NADV or DAD_NSOL received through the
      BINDING_ANCHOR port is discarded.



Bagnulo & Garcia-Martinez    Standards Track                   [Page 26]

RFC 7219                        SEND SAVI                       May 2014


   Messages received from a Validating port different from the
   BINDING_ANCHOR:

   o  If a validated DAD_NSOL message is received through port VP'
      different from the BINDING_ANCHOR port, it is forwarded to the
      appropriate Trusted ports, the LIFETIME is set to TENT_LT, the
      BINDING_ANCHOR is set to VP', and the state is changed to
      TENTATIVE_DAD.

   o  Any packet other than validated DAD_NSOL received through port VP'
      MUST NOT be forwarded unless the next state for the binding is
      VALID.  The packets received MAY be discarded or MAY be stored to
      be sent if the state changes later to VALID.  The state is left
      unchanged.

   Messages received from a Trusted port:

   o  If a DAD_NSOL message is received through a Trusted port, it is
      forwarded to the BINDING_ANCHOR port, and the state is left
      unchanged.

   o  Any other packet received from a Trusted port is forwarded
      appropriately.  This packet may come from a SEND SAVI device that
      has securely validated the attachment of the node to its
      Validating port, according to SEND SAVI rules.  The state is left
      unchanged.

   LIFETIME expires:

   o  If LIFETIME expires, the LIFETIME is cleared and the state is
      changed to NO_BIND.

3.4.  SEND SAVI Port Configuration Guidelines



   The detailed guidelines for port configuration in SEND SAVI devices
   are:

   o  Ports connected to another SEND SAVI device MUST be configured as
      Trusted ports.  Not doing so will prevent off-link traffic from
      being forwarded, along with the following effects for on-link
      traffic: significantly increase the CPU time, memory consumption,
      and signaling traffic due to SEND SAVI validation, in both the
      SEND SAVI devices and the node whose address is being validated.

   o  Ports connected to hosts SHOULD be configured as Validating ports.
      Not doing so will allow the host connected to that port to send
      packets with a spoofed source address.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 27]

RFC 7219                        SEND SAVI                       May 2014


   o  No more than one host SHOULD be connected to each port.
      Connecting more than one host to a port will allow hosts to
      generate packets with the same source address as the other hosts
      connected to the same port, and will allow replaying attacks to be
      performed as described in Section 5.1.

   o  Ports connected to routers MUST be configured as Trusted ports.
      Not doing so results in SEND SAVI devices discarding off-link
      traffic.  Note that this means that since routers are connected
      through Trusted ports, they can generate traffic with any source
      address, even those belonging to the link.

   o  Ports connected to a chain of one or more legacy switches that
      have other SEND SAVI devices but have no routers or hosts attached
      to them SHOULD be configured as Trusted ports.  Not doing so will
      significantly increase the memory consumption in the SEND SAVI
      devices and increase the signaling traffic due to SEND SAVI
      validation.

3.5.  VLAN Support



   In the case where the SEND SAVI device is a switch that supports
   customer VLANs [IEEE.802-1Q.2005], the SEND SAVI specification MUST
   behave as if there was one SEND SAVI process per customer VLAN.  The
   SEND SAVI process of each customer VLAN will store the binding
   information corresponding to the nodes attached to that particular
   customer VLAN.

3.6.  Protocol Constants



   TENT_LT is 500 milliseconds.

   DEFAULT_LT is 5 minutes.


















Bagnulo & Garcia-Martinez    Standards Track                   [Page 28]

RFC 7219                        SEND SAVI                       May 2014


4.  Protocol Walk-Through



   In this section, we include two cases that illustrate the behavior of
   SEND SAVI, the change of the attachment port of a host, and the
   attack of a malicious host.  We use the topology depicted in
   Figure 4.
               +---+
               | H |
               +---+
                 |
                 |
               +-1-----2-+       +-1-----2-+
               |         |       |         |
               |  SAVI1  |       |  SAVI2  |
               |         |       |         |
               +-3-----4-+       +-3-----4-+
                 |                 |
                 -------------------

     Figure 4: Reference SEND SAVI Topology for Protocol Walk-Through

4.1.  Change of the Attachment Point of a Host



   There are two cases, depending on whether the host H moves to a
   different port on the same switch or to a different switch.

4.1.1.  Moving to a Port of the Same Switch



   Host H is connected to port 1 of SAVI1 and moves to port 2 of the
   same switch.  Before moving, the SEND SAVI state associated to IPH,
   the IP address of H, is:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND

   In the general case, H issues a DAD_NSOL message for IPH when it is
   connected to a different port.  When SAVI1 receives this message, it
   validates it and changes its state to:

   SAVI1=TESTING_VP', BINDING_ANCHOR=1, ALTERNATIVE_BINDING_ANCHOR=2,
   TIMER=TENT_LT / SAVI2=NO_BIND

   The DAD_NSOL message is propagated to port 1, because it is the
   current BINDING_ANCHOR, and the Trusted port 3; it is not propagated
   to Validating port 4.  SAVI1 configures a timer for TENT_LT seconds.
   In addition, SAVI1 generates a NUD_NSOL and sends it through port 1.
   When SAVI2 receives this message through its Trusted port, it
   discards it and remains in the NO_BIND state.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 29]

RFC 7219                        SEND SAVI                       May 2014


   SAVI1 waits for a NUD_NADV message to be received from port 1.  Since
   there is no node attached to 1, there is no response for either of
   these messages.  When TENT_LT expires at SAVI1, the state changes to:

   SAVI1=VALID, BINDING_ANCHOR=2 / SAVI2=NO_BIND

   If the node moving does not issue a DAD_NSOL when it attaches to port
   2, then SAVI1 will receive a data packet through this port.  The data
   packet is discarded, SAVI1 issues a secured NUD_NSOL through port 1,
   and the state changes to TESTING_VP'.

   SAVI1=TESTING_VP', BINDING_ANCHOR=1, ALTERNATIVE_BINDING_ANCHOR=2
   TIMER=TENT_LT / SAVI2=NO_BIND

   SAVI1 waits for a NUD_NADV message to be received from port 1.  Since
   there is no node attached to 1, there is no response for neither of
   these messages.  When TENT_LT expires at SAVI1, the state changes to:

   SAVI1=VALID, BINDING_ANCHOR=2 / SAVI2=NO_BIND

   An alternative behavior allowed by the specification for the case in
   which the host does not issue a DAD_NSOL is that SAVI1 does nothing.
   In this case, after some time (bounded by DEFAULT_LT), the switch
   will change the state for IPH to TESTING_VP, check if H is still at
   port 1 (which it is not), and move the state to NO_BIND.  Then, a
   packet arriving from port 2 would trigger a process that finishes
   with a VALID stated with BINDING_ANCHOR=2.

4.1.2.  Moving to a Port of a Different Switch



   Host H, connected to port 1 of SAVI1, moves to port 4 of SAVI2.
   Before moving, the SEND SAVI state associated to IPH, the IP address
   of H, is:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND

   If H issues a DAD_NSOL message for IPH when it connects to port 4 of
   SAVI2, the state is changed to:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=TENTATIVE_DAD,
   BINDING_ANCHOR=4, TIMER=TENT_LT

   The DAD_NSOL message is propagated only through the Trusted port of
   SAVI2.  Then, SAVI1 changes its state as follows:

   SAVI1=TESTING_VP, BINDING_ANCHOR=1, TIMER=TENT_LT /
   SAVI2=TENTATIVE_DAD, BINDING_ANCHOR=4, TIMER=TENT_LT




Bagnulo & Garcia-Martinez    Standards Track                   [Page 30]

RFC 7219                        SEND SAVI                       May 2014


   SAVI1 propagates the DAD_NSOL message to port 1.  Since the only node
   that can answer with a secured DAD_NUD has moved, the timer at SAVI2
   expires, and SAVI2 changes its state to VALID:

   SAVI1=TESTING_VP, BINDING_ANCHOR=1, TIMER=TENT_LT / SAVI2=VALID,
   BINDING_ANCHOR=4

   Just a very short time after, the timer at SAVI1 expires, and the
   state changes to NO_BIND:

   SAVI1=NO_BIND / SAVI2=VALID, BINDING_ANCHOR=4

   If host H does not send a DAD_NSOL when it moves to SAVI2 but instead
   sends a data packet, SAVI2 changes its state to TENTATIVE_NUD:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=TENTATIVE_NUD,
   BINDING_ANCHOR=4, TIMER=TENT_LT

   SAVI2 issues a secured NUD_NSOL through port 4.  H is assumed to have
   the address configured (otherwise, it should not have generated a
   data packet), so it can respond with a NUD_NADV.  When SAVI1 receives
   the NUD_NADV and validates it, the state is changed to VALID:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=VALID, BINDING_ANCHOR=4

   After some time (bounded by DEFAULT_LT), the state in SAVI1 will
   expire, and SAVI1 will perform a check for host H:

   SAVI1=TESTING_VP, BINDING_ANCHOR=1, TIMER=TENT_LT / SAVI2=VALID,
   BINDING_ANCHOR=4

   SAVI1 issues a NUD_NSOL through port 1 for IPH.  No response is
   received in this case, so SAVI1 changes its state to NO_BIND:

   SAVI1=NO_BIND / SAVI2=VALID, BINDING_ANCHOR=4

4.2.  Attack of a Malicious Host



   Host H is attached to the SEND SAVI infrastructure through port 1 of
   SAVI1.  We consider that host M starts sending data packets using IPH
   (the IP address of H) as the source address, without issuing a
   DAD_NSOL (a similar analysis can be done for this case).

4.2.1.  M Attaches to the Same Switch as the Victim's Switch



   The initial state before the attack of M is:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND



Bagnulo & Garcia-Martinez    Standards Track                   [Page 31]

RFC 7219                        SEND SAVI                       May 2014


   M attaches to port 2 of SAVI1 and starts sending data packets.  When
   SAVI1 receives the data packet, the packet is discarded.  SEND SAVI
   may issue a secured NUD_NSOL through port 1 and changes the state to:

   SAVI1=TESTING_VP', BINDING_ANCHOR=1, ALTERNATIVE_BINDING_ANCHOR=2,
   TIMER=TENT_LT / SAVI2=NO_BIND

   Host H is still attached to port 1, so it receives the NUD_NSOL and
   responds with a secured NUD_NADV.  SAVI1 receives this message,
   validates it, and changes its state again to:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND

   To prevent the drain of CPU resources in SAVI1, the processing of
   further packets received from port 2 may be rate-limited, as
   discussed in Section 5.2.

   An alternative to the previous behavior is that SAVI1 does nothing
   when node M starts sending packets from port 2.  In this case, when
   the timer to renew the state triggers (this time it's bounded by
   DEFAULT_LT), SAVI1 moves the state to TESTING_VP, sends a NUD_NSOL
   through port 1, host H responds, and the state remains in VALID for
   BINDING_ANCHOR=1.  In this way, communication of host H is also
   defended.

4.2.2.  M Attaches to a Different Switch to the Victim's Switch



   The initial state before the attack of M is:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND

   M attaches to port 2 of SAVI2 and starts sending data packets.  When
   SAVI2 receives the data packet, it changes the state to:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=TENTATIVE_DAD,
   BINDING_ANCHOR=2, TIMER=TENT_LT

   SAVI2 issues a secured NUD_NSOL through port 2.  Since M does not own
   the IPH CGA, it cannot respond to the message.  When the timer
   expires, the state is moved back to:

   SAVI1=VALID, BINDING_ANCHOR=1 / SAVI2=NO_BIND

   To prevent the drain of CPU resources in SAVI2, the processing of
   further packets received from port 2 may be rate-limited, as
   discussed in Section 5.2.





Bagnulo & Garcia-Martinez    Standards Track                   [Page 32]

RFC 7219                        SEND SAVI                       May 2014


5.  Security Considerations



   SEND SAVI operates only with validated SEND messages to create
   bindings.  Note that IPv6 packets generated by non-SEND nodes will be
   discarded by the first SEND SAVI device receiving it.  Therefore,
   attackers cannot obtain any benefit by not using SEND.  In order to
   perform address validation in a mixed scenario comprising SEND and
   non-SEND devices, a different solution is required, which should be
   addressed in another document.

   Nodes MUST NOT assume that all SEND messages received from a SEND
   SAVI device are validated, since these devices only validate the
   messages strictly required for SEND SAVI operation.  Among the number
   of messages that are not validated by SEND SAVI, we can name NUD_NSOL
   messages generated by other nodes and its corresponding NUD_NADV
   responses, or RSOL messages.

   SEND SAVI improves protection compared to conventional SAVI as a
   result of the increased ability of SEND nodes to prove address
   ownership.

   A critical security consideration regarding SEND SAVI deals with the
   need of proper configuration of the roles of the ports in a SEND SAVI
   deployment scenario.  Regarding security, the main requirement is
   that ports defining the protected perimeter SHOULD be configured as
   Validating ports.  Not doing so will allow an attacker to send
   packets using any source address, regardless of the bindings
   established in other SEND SAVI devices.

5.1.  Protection against Replay Attacks



   One possible concern about SEND SAVI is its behavior when an attacker
   tries to forge the identity of a legitimate node by replaying SEND
   messages used by the SEND SAVI specification.  An attacker could
   replay any of these messages to interfere with the SEND SAVI
   operation.  For example, it could replay a DAD_NSOL message to abort
   the configuration of an address for a legitimate node and to gain the
   right to use the address for DEFAULT_LT seconds.

   We can analyze two different cases when considering SEND SAVI replay
   attacks:

   o  When the SEND message replayed is used to create or update binding
      information for SEND SAVI, since the port through which this
      message is received is key to the SEND SAVI operation.  SEND SAVI
      creates and maintains bindings as a result of the reception of
      DAD_NSOL messages and of the exchange of NUD_NSOL/NUD_NADV
      messages.



Bagnulo & Garcia-Martinez    Standards Track                   [Page 33]

RFC 7219                        SEND SAVI                       May 2014


   o  When the SEND message replayed does not result in the update of
      binding information for SEND SAVI and, thus, is not related to the
      specific port through which it was received.  Such situations are
      the reception of CPA messages containing certificates, and the
      processing of an RADV message coming from a Trusted port, which
      can be used in SEND SAVI to populate the SEND SAVI Prefix List.
      In these two cases, the security risks are equivalent to those of
      the SEND operation, i.e., we can consider that the information
      will not be changed by its legitimate sender for the time during
      which the SEND specification allows replaying (which depends on
      the values of TIMESTAMP_FUZZ and TIMESTAMP_DRIFT [RFC3971]).

   For replay of messages belonging to the second case, i.e., messages
   that do not result in changes in the SEND SAVI binding information,
   the security provided by SEND is sufficient.  For the replay of
   messages belonging to the first case, DAD_NSOL and NUD_NSOL/NUD_NADV
   messages, protection results from the behavior of SEND SAVI, as
   specified in Section 3.3.2, which restricts the ports to which the
   messages involved in SEND SAVI binding updates are disseminated.
   SEND SAVI devices only forward these messages to ports for which a
   binding to the address being tested by the DAD_NSOL message existed.
   Therefore, it is not enough for an attacker to subscribe to a
   Solicited Node address to receive DAD_NSOL messages sent to that
   address, but the attacker needs to generate a valid DAD_NSOL message
   associated to the address for which the binding is being tested,
   which is deemed unfeasible [RFC3971].

5.2.  Protection against Denial-of-Service Attacks



   The attacks against the SEND SAVI device basically consist of making
   the SEND SAVI device consume its resources until it runs out of them.
   For instance, a possible attack would be to send packets with
   different source addresses, making the SEND SAVI device create state
   for each of the addresses and waste memory.  At some point, the SEND
   SAVI device runs out of memory and needs to decide how to react.  The
   result is that some form of garbage collection is needed to prune the
   entries.  When the SEND SAVI device runs out of the memory allocated
   for the SEND SAVI Database, it is RECOMMENDED that it creates new
   entries by deleting the entries with a higher Creation time.  This
   implies that older entries are preserved and newer entries overwrite
   each other.  In an attack scenario where the attacker sends a batch
   of data packets with different source addresses, each new source
   address is likely to rewrite another source address created by the
   attack itself.  It should be noted that entries are also garbage
   collected using the DEFAULT_LT, which is updated by NUD_NSOL/NUD_NADV
   exchanges.  The result is that in order for an attacker to actually
   fill the SEND SAVI Database with false source addresses, it needs to
   continuously answer to NUD_NSOL for all the different source



Bagnulo & Garcia-Martinez    Standards Track                   [Page 34]

RFC 7219                        SEND SAVI                       May 2014


   addresses, so that the entries grow old and compete with the
   legitimate entries.  The result is that the cost of the attack is
   highly increased for the attacker.

   In addition, it is also RECOMMENDED that a SEND SAVI device reserves
   a minimum amount of memory for each available port (in the case where
   the port is used as part of the L2 anchor).  The REQUIRED minimum is
   the memory needed to store four bindings associated to the port,
   although it SHOULD be raised if the ratio between the maximum number
   of bindings allowed in the device and the number of ports is high.
   The motivation for setting a minimum number of bindings per port is
   as follows.  An attacker attached to a given port of a SEND SAVI
   device may attempt to launch a DoS attack towards the SEND SAVI
   device by creating many bindings for different addresses.  It can do
   so by sending DAD_NSOL for different addresses.  The result is that
   the attack will consume all the memory available in the SEND SAVI
   device.  The above recommendation aims to reserve a minimum amount of
   memory per port, so that nodes located in different ports can make
   use of the reserved memory for their port even if a DoS attack is
   occurring in a different port.

   The SEND SAVI device may store data packets while the address is
   being verified, for example, when a DAD_NSOL is lost before arriving
   to the SEND SAVI device to which the host attaches; when the host
   sends data packets, these data packets may be stored until the SEND
   SAVI device verifies the binding by means of a NUD packet exchange.
   In this case, the memory for data packet storage may also be a target
   of DoS attacks.  A SEND SAVI device MUST limit the amount of memory
   used to store data packets, allowing the other functions (such as
   being able to store new bindings) to have available memory even in
   the case of an attack, such as those described above.

   It is worth noting that the potential of DoS attacks against the SEND
   SAVI network is increased due to the use of costly cryptographic
   operations in order to validate the address of the nodes.  An
   attacker could generate packets using new source addresses in order
   to make the closest SEND SAVI device spend CPU time to validate
   DAD_NSOL messages or to generate a secure NUD_NSOL.  This attack can
   be used to drain CPU resources of SEND SAVI devices with a very low
   cost for the attacker.  In order to solve this problem, rate-limiting
   the processing of packets that trigger SEND SAVI events SHOULD be
   enforced on a per-port basis.









Bagnulo & Garcia-Martinez    Standards Track                   [Page 35]

RFC 7219                        SEND SAVI                       May 2014


5.3.  Considerations on the Deployment Model for Trust Anchors



   The SEND specification [RFC3971] proposes two deployment models for
   trust anchors: either a centralized model relaying on a globally
   rooted public key infrastructure or a more local, decentralized
   deployment model in which end hosts are configured with a collection
   of public keys that are trusted only on a domain.

   The appeal of a centralized model is the possibility for hosts to use
   SEND to validate routers as they move through links belonging to
   different organizations without additional configuration.  However,
   without any further protection, it also enables routers authorized
   with a certificate path rooted on a global trust anchor to appear as
   legitimate routers in a link in which they were not intended to act
   as such.  This threat already existed for SEND deployments, for which
   links configured to accept centralized trust anchors may send
   outgoing traffic and use prefix information from alien routers.  In a
   SEND SAVI deployment, such routers may be able to deliver off-link
   traffic to any node of the link.

   In order to cope with this threat, SEND SAVI specifies that nodes are
   only allowed to behave as routers if they connect through Trusted
   ports.  In particular, RADV messages and traffic with off-link source
   addresses are discarded when received through Validating ports, which
   are the ports intended for non-trusted infrastructure, as moving
   nodes.  The protection provided by filtering RADV messages prevents
   SEND nodes from identifying alien routers as legitimate routers, even
   though the trust anchor of these routers is valid.

   Besides, it is worth to say that SEND SAVI supports a decentralized
   deployment model.

5.4.  Residual Threats



   SEND SAVI assumes that a host will be able to defend its address when
   the DAD procedure is executed for its addresses, and that it will
   answer to a NUD_NSOL with a NUD_NADV when required.  This is needed,
   among other things, to support mobility within a link (i.e., to allow
   a host to detach and reconnect to a different layer-2 anchor of the
   same IP subnetwork, without changing its IP address).  If the SEND
   SAVI device does not see the DAD_NADV or the NUD_NADV, it may grant
   the binding to a different binding anchor.  This means that if an
   attacker manages to prevent a host from defending its source address,
   it will be able to destroy the existing binding and create a new one,
   with a different binding anchor.  An attacker may do so, for example,
   by launching a DoS attack to the host that will prevent it to issue
   proper replies.




Bagnulo & Garcia-Martinez    Standards Track                   [Page 36]

RFC 7219                        SEND SAVI                       May 2014


5.5.  Privacy Considerations



   A SEND SAVI device MUST delete binding anchor information as soon as
   possible (i.e., as soon as the state for a given address is back to
   NO_BIND), except where there is an identified reason why that
   information is likely to be involved in the detection, prevention, or
   tracing of actual source address spoofing.  Information about the
   majority of hosts that never spoof SHOULD NOT be logged.

6.  Acknowledgments



   Thanks to Jean-Michel Combes, Ana Kukec, Ted Lemon, Adrian Farrel,
   Barry Leiba, Brian Haberman, Vicent Roca, and Benoit Claise for their
   reviews and comments on this document.  The text has also benefited
   from feedback provided by Tony Cheneau and Greg Daley.

   Marcelo Bagnulo is partly funded by Trilogy 2, a research project
   supported by the European Commission under its Seventh Framework
   Program.  Alberto Garcia-Martinez was supported, in part, by project
   TEC2012-38362-C03-01, granted by the Spanish Economy and
   Competitiveness Ministry.

7.  References



7.1.  Normative References



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

   [RFC3971]  Arkko, J., Kempf, J., Zill, B., and P. Nikander, "SEcure
              Neighbor Discovery (SEND)", RFC 3971, March 2005.

   [RFC3972]  Aura, T., "Cryptographically Generated Addresses (CGA)",
              RFC 3972, March 2005.

   [RFC4861]  Narten, T., Nordmark, E., Simpson, W., and H. Soliman,
              "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861,
              September 2007.

   [RFC4862]  Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless
              Address Autoconfiguration", RFC 4862, September 2007.










Bagnulo & Garcia-Martinez    Standards Track                   [Page 37]

RFC 7219                        SEND SAVI                       May 2014


7.2.  Informative References



   [IEEE.802-1Q.2005]
              Institute of Electrical and Electronics Engineers, "IEEE
              Standard for Local and Metropolitan Area Networks /
              Virtual Bridged Local Area Networks", IEEE Standard
              802.1Q, May 2005.

   [RFC2827]  Ferguson, P. and D. Senie, "Network Ingress Filtering:
              Defeating Denial of Service Attacks which employ IP Source
              Address Spoofing", BCP 38, RFC 2827, May 2000.

   [RFC6434]  Jankiewicz, E., Loughney, J., and T. Narten, "IPv6 Node
              Requirements", RFC 6434, December 2011.

   [RFC6620]  Nordmark, E., Bagnulo, M., and E. Levy-Abegnoli, "FCFS
              SAVI: First-Come, First-Served Source Address Validation
              Improvement for Locally Assigned IPv6 Addresses", RFC
              6620, May 2012.

   [RFC7039]  Wu, J., Bi, J., Bagnulo, M., Baker, F., and C. Vogt,
              "Source Address Validation Improvement (SAVI) Framework",
              RFC 7039, October 2013.

Authors' Addresses



   Marcelo Bagnulo
   Universidad Carlos III de Madrid
   Av. Universidad 30
   Leganes, Madrid  28911
   Spain

   Phone: 34 91 6248814
   EMail: marcelo@it.uc3m.es
   URI:   http://www.it.uc3m.es


   Alberto Garcia-Martinez
   Universidad Carlos III de Madrid
   Av. Universidad 30
   Leganes, Madrid  28911
   Spain

   Phone: 34 91 6248782
   EMail: alberto@it.uc3m.es
   URI:   http://www.it.uc3m.es





Bagnulo & Garcia-Martinez    Standards Track                   [Page 38]