Network Working Group                                  Sina Mirtorabi
Internet Draft                                         Abhay Roy
Document: draft-ietf-ospf-af-alt-00.txt                Michael Barnes
Expiration Date: September 2004                        Cisco Systems

                                                       Acee Lindem
                                                       Redback Networks

                                                       Quaizar Vohra
                                                       Rahul Aggarwal
                                                       Juniper Networks

                                                       March 2004

                     Support of address families in OSPFv3

Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026.

   Internet Drafts are working documents of the Internet Engineering
   Task Force (IETF), its Areas, and its Working Groups. Note that other
   groups may also distribute working documents as Internet Drafts.

   Internet Drafts are draft documents valid for a maximum of six
   months. Internet Drafts may be updated, replaced, or obsoleted by
   other documents at any time. It is not appropriate to use Internet
   Drafts as reference material or to cite them other than as a "working
   draft" or "work in progress".

   The list of current Internet-Drafts can be accessed at

   The list of Internet-Draft Shadow Directories can be accessed at


   This document describes a mechanism for supporting multiple address
   families in OSPFv3 using multiple instances. It maps an address
   family (AF) to an OSPFv3 instance using the Instance ID field in the
   OSPFv3 packet header. This approach is fairly simple and minimizes
   extensions to OSPFv3 for supporting multiple AF's.

Mirtorabi et al.                                                 [Page 1]

Internet Draft          Address Families in OSPFv3             March 2003

1. Motivation

   OSPFv3 has been defined to support IPv6 unicast AF. There is a need
   to carry other AFs in OSPFv3 such as multicast IPv6, unicast or
   multicast IPv4. This document introduces these other AFs in OSPFv3
   by reserving Instance IDs and using one OSPFv3 instance for one AF.

2. Proposed Solution

   Currently the entire Instance ID number space is used for IPv6
   unicast. We propose to assign different ranges to different AF's in
   order to support other AF's in OSPFv3. Each AF will establish
   different adjacency, have different link state database and compute
   different shortest path tree. Additionally, the current LSAs that are
   defined to carry IPv6 unicast prefix can be used without any
   modification in different instances to carry different AF's prefixes.

   It should be noted that OSPFv3 is running on the top of IPv6 and uses
   IPv6 link local address for OSPFv3 control packet and next hop
   calculation. Therefore, it is required that IPv6 be enabled on a link,
   although the link may not be participating in IPv6 unicast AF.

3. Instance ID values for new AF's

   Instance ID zero is already used by default for IPv6 unicast AF.
   We define the following ranges for different AF's. The first value
   of each range is considered as the default value for the
   corresponding AF.

   Instance ID # 0    -  # 31     IPv6 unicast AF
   Instance ID # 32   -  # 63     IPv6 multicast AF
   Instance ID # 64   -  # 95     IPv4 unicast AF
   Instance ID # 96   -  # 127    IPv4 multicast AF
   Instance ID # 128  -  # 255    Reserved

4. New bit in Options field

   A new bit is defined in the Options field for AF support.

         0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6   7  8  9  0  1  2  3
         | | | | | | | | | | | | | | | | | AF|DC| R| N|MC| E|V6|

Mirtorabi et al.                                                 [Page 2]

Internet Draft          Address Families in OSPFv3             March 2003


    When a router supports AF, it MUST set this bit in the Options
    field of Hello Packets, DD packets and LSAs.

5. Changes to the Hello processing

   When a router does not support an AF but it is configured with an
   Instance ID in the same range, packets could be blackholed. This
   could happen due to misconfiguration or router downgrade to a
   previous code level. Blackholing is possible because the router which
   doesn't support the AF can still be included in the SPF calculated
   path as long as it establishes adjacencies using the Instance ID
   corresponding to the AF. Note that router and network LSAs are AF

   In order to avoid the above situation, hello processing is changed in
   order to only establish adjacency with the routers that have the
   AF-bit set in their Options field.

   Receiving Hello Packets is specified in section of [Ref1].
   The following check is added to Hello reception:

   When a router participate in an AF (sets the AF-bit in Options field)
   it MUST discard Hello packets having the AF-bit clear in the Options
   field. The only exception is IPv6 unicast AF, where this check MUST
   NOT be done (to help backward compatibility).

6. Modification to some of the bits defined in [Ref1]

   Some of the bits defined in OSPFv3 are relevant to IPv6 unicast
   AF, and are not needed in other AF's. Some may be applicable only
   to a certain AF. Below is the list of changes to those bits:

   o Options Field

        0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8  9  0  1  2  3
         | | | | | | | | | | | | | | | | | |DC| R| N|* | E|* |

      o V6-bit

        The V6 bit is used in OSPFv3 to exclude a node from IPv6 unicast
        route calculation but allow it in the SPF calculation for
        other address families. Since Instance ID now denotes the AF
        explicitly, this bit is ignored in AF's other than IPv6 unicast.

Mirtorabi et al.                                                 [Page 3]

Internet Draft          Address Families in OSPFv3             March 2003

      o MC-bit

        This bit is not used in other AF's introduced in this document.

   o Prefix Options Field

             0  1  2  3  4  5  6  7
            |  |  |  |  | P|* |LA|NU|

      o MC bit in the Prefix Options field

        This bit is not used in other AF's introduced in this document.

      o NU bit usage in the Prefix Options field

        The NU bit must be clear in all unicast AF's and it must be set
        in all multicast AF's.

   Note that all bits unused in a given AF could be redefined later.

7. Carrying Prefixes in new AF's

   Each Prefix defined in OSPFv3 has a prefix length field. This
   facilitate advertising prefixes of different lengths in different
   AF's. The existing LSAs defined in OSPFv3 are used for this
   purpose and there is no need to define new LSAs.

8. Next hop for IPv4 unicast and multicast AF's

   OSPFv3 runs on the top of IPv6 and uses IPv6 link local addresses
   for OSPFv3 control packets and next hop calculations. Although IPV6
   link local addresses could be used as next hops for IPv4 address
   families, it is desirable to have IPv4 next hop addresses. For
   example, in IPv4 multicast having the nexthop address the same as
   the PIM neighbor address (IPv4 address) makes it easier to know to
   which upstream neighbor to send a PIM join when doing a RPF lookup
   for a source. It is also easier for troubleshooting purposes to have
   a next hop with the same semantics as the AF.

   In order to achieve this, the link's IPv4 address will be advertised
   in the "link local address" field of the IPv4 instance's Link-LSA.
   This address is placed in the first 32 bit of "link local address"
   field and used for IPv4 next hop calculations.

Mirtorabi et al.                                                 [Page 4]

Internet Draft          Address Families in OSPFv3             March 2003

   We call direct interface address (DIA) the address that is reachable
   directly via the link provided that a layer 3 to layer 2 mapping is
   available. Note that there is no explicit need for the IPv4 link
   addresses to be on the same subnet. An implementation should resolve
   layer 3 to layer 2 mappings via ARP or ND for a DIA even if the IPv4
   address is not on the same subnet as the router's interface IP address.

9. OSPFv3 over IPv4

   Although OSPFv3 is defined to run over IPv6, it is possible to run
   OSPFv3 over IPv4 and making IPv4 address-family, IPv6 independent.
   This is achieved by using IPv4 for OSPFv3 control packet with the
   same protocol number 89 as OSPFv2. The version in the OSPF packet
   allows to distinguish between OSPFv2 and OSPFv3.

   We define a parameter in interface data structure called V4TransProtocol.
   V4TransProtocol flag can have two values: Enabled or Disabled, the default
   value being Disabled.

   when set to Enabled OSPFv3 use IPv4 as transport protocol
   When set to Disabled OSPFv3 use IPv6 as transport protocol

   Note that this parameter can only be enabled in IPv4 address family
   (see section 3 for instance id range).

10. Virtual Link (VL)

   When OSPFv3 is running over IPv4 there is no special requirement for
   VL to be operational since OSPFv3 control packet are sent over IPv4.
   However note that all routers within the area should run over IPv4
   or provided that there is a contiguous path between the virtual link
   end point that run over IPv4.

   When OSPFv3 is running over IPv6, the control packet sent for virtual
   link are IPv6 packets and may traverse multiples hops therefore there
   must be a global IPv6 address associated with the virtual link so that
   the control packet is forwarded correctly by the intermediate hops
   between VL end points. Although this requirement can be satisfied in
   IPv6 unicast AF, this will not function in other AF as there cannot
   be a multihop forwarding based on global IPv6 address or such a path
   may not exist. Therefore virtual link are not currently supported in
   other AF's.

Mirtorabi et al.                                                 [Page 5]

Internet Draft          Address Families in OSPFv3             March 2003

11. Backward compatibility issues

   Each new AF will have their corresponding Instance ID and can
   operate with the existing non-capable routers in IPv6 unicast
   topology. Further, when a non-capable router uses an Instance ID which
   is reserved for a given AF, since the non-capable router will not have
   the AF-bit set in the Hello an adjacency will not be established with
   an AF capable router. Therefore, there are no backward compatibility
   issues. AF's can be gradually deployed without disturbing networks with
   current non-capable routers.

12. Address-family design Considerations

   This section describes the rationale for adopting the multiple
   instance ID approach for supporting multiple address families in
   OSPFv3. As described earlier, OSPFv3 is designed to support multiple
   instances. Hence mapping an instance to an address family doesn't
   introduce new mechanisms in the protocol. It minimizes the protocol
   extensions required and it simplifies the implementation. The
   presence of a separate link state database per address family is
   also easier to debug and operate. Additionally, it doesn't change
   the existing instance, area and interface based configuration model
   in most OSPF implementations.

13. Security Considerations

   The technique described in this document does not introduce any new
   security issues to the OSPFv3 protocol.

14. References

  [Ref1] R. Coltun, D. Ferguson and J.  Moy,  "OSPF  for  IPv6",
         RFC 2740, December 1999.

Mirtorabi et al.                                                 [Page 6]

Internet Draft          Address Families in OSPFv3             March 2003

15. Authors address

   Sina Mirtorabi                          Acee Lindem
   Cisco Systems                           Redback Networks
   170 W. Tasman Dr.                       102 Carric Bend Court
   San Jose, CA 95134                      Cary, NC 27519
   Email:                   Email:

   Abhay Roy                               Quaizar Vohra
   Cisco Systems                           Juniper Networks
   170 W. Tasman Dr.                       1194 North Mathilda Ave.
   San Jose, CA 95134                      Sunnyvale, CA 94089
   Email:                    Email:

   Michael Barnes                          Rahul Aggarwal
   Cisco Systems                           Juniper Networks
   170 W. Tasman Dr.                       1194 North Mathilda Ave.
   San Jose, CA 95134                      Sunnyvale, CA 94089
   Email:               Email: