Network Working Group                                             D. Kim
Request for Comments: 3446                                         Verio
Category: Informational                                         D. Meyer
                                                              H. Kilmer
                                                           D. Farinacci
                                                       Procket Networks
                                                           January 2003


            Anycast Rendevous Point (RP) mechanism using
                Protocol Independent Multicast (PIM)
            and Multicast Source Discovery Protocol (MSDP)

Status of this Memo

  This memo provides information for the Internet community.  It does
  not specify an Internet standard of any kind.  Distribution of this
  memo is unlimited.

Copyright Notice

  Copyright (C) The Internet Society (2003).  All Rights Reserved.

Abstract

  This document describes a mechanism to allow for an arbitrary number
  of Rendevous Points (RPs) per group in a single shared-tree Protocol
  Independent Multicast-Sparse Mode (PIM-SM) domain.

1. Introduction

  PIM-SM, as defined in RFC 2362, allows for only a single active RP
  per group, and as such the decision of optimal RP placement can
  become problematic for a multi-regional network deploying PIM-SM.

  Anycast RP relaxes an important constraint in PIM-SM, namely, that
  there can be only one group to RP mapping can be active at any time.
  The single mapping property has several implications, including
  traffic concentration, lack of scalable register decapsulation (when
  using the shared tree), slow convergence when an active RP fails,
  possible sub-optimal forwarding of multicast packets, and distant RP
  dependencies.  These properties of PIM-SM have been demonstrated in
  native continental or inter-continental scale multicast deployments.
  As a result, it is clear that ISP backbones require a mechanism that
  allows definition of multiple active RPs per group in a single PIM-SM
  domain.  Further, any such mechanism should also address the issues
  addressed above.




Kim, et al.                  Informational                      [Page 1]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


  The mechanism described here is intended to address the need for
  better fail-over (convergence time) and sharing of the register
  decapsulation load (again, when using the shared-tree) among RPs in a
  domain.  It is primarily intended for applications within those
  networks using MBGP, Multicast Source Discovery Protocol [MSDP] and
  PIM-SM protocols, for native multicast deployment, although it is not
  limited to those protocols.  In particular, Anycast RP is applicable
  in any PIM-SM network that also supports MSDP (MSDP is required so
  that the various RPs in the domain maintain a consistent view of the
  sources that are active).  Note however, a domain deploying Anycast
  RP is not required to run MBGP.  Finally, a general requirement of
  the Anycast RP scheme is that the anycast address MUST NOT be used as
  the RP address in the RP's SA messages.

  The keywords MUST, MUST NOT, MAY, OPTIONAL, REQUIRED, RECOMMENDED,
  SHALL, SHALL NOT, SHOULD, SHOULD NOT are to be interpreted as defined
  in BCP 14, RFC 2119 [RFC2119].

2. Problem Definition

  The anycast RP solution provides a solution for both fast fail-over
  and shared-tree load balancing among any number of active RPs in a
  domain.

2.1. Traffic Concentration and Distributing Decapsulation Load Among RPs

  While PIM-SM allows for multiple RPs to be defined for a given group,
  only one group to RP mapping can be active at a given time.  A
  traditional deployment mechanism for balancing register decapsulation
  load between multiple RPs covering the multicast group space is to
  split up the 224.0.0.0/4 space between multiple defined RPs.  This is
  an acceptable solution as long as multicast traffic remains low, but
  has problems as multicast traffic increases, especially because the
  network operator defining group space split between RPs does not
  always have a priori knowledge of traffic distribution between
  groups.  This can be overcome via periodic reconfigurations, but
  operational considerations cause this type of solution to scale
  poorly.

2.2. Sub-optimal Forwarding of Multicast Packets

  When a single RP serves a given multicast group, all joins to that
  group will be sent to that RP regardless of the topological distance
  between the RP and the sources and receivers.  Initial data will be
  sent towards the RP also until configured the shortest path tree
  switch threshold is reached, or the data will always be sent towards
  the RP if the network is configured to always use the RP rooted
  shared tree.  This holds true even if all the sources and the



Kim, et al.                  Informational                      [Page 2]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


  receivers are in any given single region, and RP is topologically
  distant from the sources and the receivers.  This is an artifact of
  the dynamic nature of multicast group members, and of the fact that
  operators may not always have a priori knowledge of the topological
  placement of the group members.

  Taken together, these effects can mean that (for example) although
  all the sources and receivers of a given group are in Europe, they
  are joining towards the RP in the USA and the data will be traversing
  a relatively expensive pipe(s) twice, once to get to RP, and back
  down the RP rooted tree again, creating inefficient use of expensive
  resources.

2.3. Distant RP Dependencies

  As outlined above, a single active RP per group may cause local
  sources and receivers to become dependent on a topologically distant
  RP.  In addition, when multiple RPs are configured, there can be
  considerable convergence delay involved in switching to the backup
  RP.  This delay may exist independent of the toplogical location of
  the primary and backup RPs.

3. Solution

  Given the problem set outlined above, a good solution would allow an
  operator to configure multiple RPs per group, and distribute those
  RPs in a topologically significant manner to the sources and
  receivers.

3.1. Mechanisms

  All the RPs serving a given group or set of groups are configured
  with an identical anycast address, using a numbered interface on the
  RPs (frequently a logical interface such as a loopback is used).  RPs
  then advertise group to RP mappings using this interface address.
  This will cause group members (senders) to join (register) towards
  the topologically closest RP.  RPs MSDP peer with each other using an
  address unique to each RP.  Since the anycast address is not a unique
  address (by definition), a router MUST NOT choose the anycast unicast
  address as the router ID, as this can prevent peerings and/or
  adjacencies from being established.

  In summary then, the following steps are required:








Kim, et al.                  Informational                      [Page 3]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


3.1.1. Create the set of group-to-anycast-RP-address mappings

  The first step is to create the set of group-to-anycast-RP-address
  mappings to be used in the domain.  Each RP participating in an
  anycast RP set must be configured with a consistent set of group to
  RP address mappings.  This mapping will be used by the non-RP routers
  in the domain.

3.1.2. Configure each RP for the group range with the anycast RP address

  The next step is to configure each RP for the group range with the
  anycast RP address.  If a dynamic mechanism, such as auto-RP or the
  PIMv2 bootstrap mechanism, is being used to advertise group to RP
  mappings, the anycast IP address should be used for the RP address.

3.1.3. Configure MSDP peerings between each of the anycast RPs in the
  set

  Unlike the group to RP mapping advertisements, MSDP peerings must use
  an IP address that is unique to the endpoints; that is, the MSDP
  peering endpoints MUST use a unicast rather than anycast address.  A
  general guideline is to follow the addressing of the BGP peerings,
  e.g., loopbacks for iBGP peering, physical interface addresses for
  eBGP peering.  Note that the anycast address MUST NOT be used as the
  RP address in SA messages (as this would case the peer-RPF check to
  fail).

3.1.4. Configure the non-RP's with the group-to-anycast-RP-address
  mappings

  Finally, each non-RP router must learn the set of group to RP
  mappings.  This could be done via static configuration, auto-RP, or
  by PIMv2 bootstrap mechanism.

3.1.5. Ensure that the anycast IP address is reachable by all routers in
  the domain

  This is typically accomplished by causing each RP to inject the /32
  into the domain's IGP.

3.2. Interaction with MSDP Peer-RPF check

  Each MSDP peer receives and forwards the message away from the RP
  address in a "peer-RPF flooding" fashion.  The notion of peer-RPF
  flooding is with respect to forwarding SA messages [MSDP].  The BGP
  routing tables are examined to determine which peer is the next hop
  towards the originating RP of the SA message.  Such a peer is called
  an "RPF peer".  See [MSDP] for details of the Peer-RPF check.



Kim, et al.                  Informational                      [Page 4]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


3.3. State Implications

  It should be noted that using MSDP in this way forces the creation of
  (S,G) state along the path from the receiver to the source.  This
  state may not be present if a single RP was used and receivers were
  forced to stay on the shared tree.

4. Security considerations

  Since the solution described here makes heavy use of anycast
  addressing, care must be taken to avoid spoofing.  In particular
  unicast routing and PIM RPs must be protected.

4.1. Unicast Routing

  Both internal and external unicast routing can be weakly protected
  with keyed MD5 [RFC1828], as implemented in an internal protocol such
  as OSPF [RFC2328] or in BGP [RFC2385].  More generally,  IPSEC
  [RFC2401] could be used to provide protocol integrity for the unicast
  routing system.

4.1.1. Effects of Unicast Routing Instability

  While not a security issue, it is worth noting that if unicast
  routing is unstable, then the actual RP that source or receiver is
  using will be subject to the same instability.

4.2. Multicast Protocol Integrity

  The mechanisms described in [RFC2362] should be used to provide
  protocol message integrity protection and group-wise message origin
  authentication.

4.3. MSDP Peer Integrity

  As is the the case for BGP, MSDP peers can be protected using keyed
  MD5 [RFC1828].

5. Acknowledgments

  John Meylor, Bill Fenner, Dave Thaler and Tom Pusateri provided
  insightful comments on earlier versions for this idea.

  This memo is a product of the MBONE Deployment Working Group (MBONED)
  in the Operations and Management Area of the Internet Engineering
  Task Force.  Submit comments to <[email protected]> or the
  authors.




Kim, et al.                  Informational                      [Page 5]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


6. References

  [MSDP]     D. Meyer and B. Fenner, Editors, "Multicast Source
             Discovery Protocol (MSDP)", Work in Progress.

  [RFC2401]  Kent, S. and R. Atkinson, "Security Architecture for the
             Internet Protocol", RFC 2401, August 1995.

  [RFC1828]  Metzger, P. and W. Simpson, "IP Authentication using Keyed
             MD5", RFC 1828, August 1995.

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

  [RFC2362]  Estrin, D., Farinacci, D., Helmy, A., Thaler, D., Deering,
             S., Handley, M., Jacobson, V., Liu, C., Sharma, P. and L.
             Wei, "Protocol Independent Multicast-Sparse Mode (PIM-SM):
             Protocol Specification", RFC 2362, June 1998.

  [RFC2328]  Moy, J., "OSPF Version 2", STD 54, RFC 2328, April 1998.

  [RFC2385]  Heffernan, A., "Protection of BGP Sessions via the TCP MD5
             Signature Option", RFC 2385, August 1998.

  [RFC2403]  Madson, C. and R. Glenn, "The Use of HMAC-MD5-96 within
             ESP and AH", RFC 2403, November 1998.

7. Author's Address

  Dorian Kim
  Verio, Inc.
  EMail: [email protected]

  Hank Kilmer
  EMail: [email protected]

  Dino Farinacci
  Procket Networks
  EMail: [email protected]

  David Meyer
  EMail: [email protected]









Kim, et al.                  Informational                      [Page 6]

RFC 3446        Anycast RP mechanism using PIM and MSDP     January 2003


8.  Full Copyright Statement

  Copyright (C) The Internet Society (2003).  All Rights Reserved.

  This document and translations of it may be copied and furnished to
  others, and derivative works that comment on or otherwise explain it
  or assist in its implementation may be prepared, copied, published
  and distributed, in whole or in part, without restriction of any
  kind, provided that the above copyright notice and this paragraph are
  included on all such copies and derivative works.  However, this
  document itself may not be modified in any way, such as by removing
  the copyright notice or references to the Internet Society or other
  Internet organizations, except as needed for the purpose of
  developing Internet standards in which case the procedures for
  copyrights defined in the Internet Standards process must be
  followed, or as required to translate it into languages other than
  English.

  The limited permissions granted above are perpetual and will not be
  revoked by the Internet Society or its successors or assigns.

  This document and the information contained herein is provided on an
  "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
  TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
  BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
  HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
  MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement

  Funding for the RFC Editor function is currently provided by the
  Internet Society.



















Kim, et al.                  Informational                      [Page 7]


You are viewing proxied material from gopher.rbfh.de. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.