Network Working Group                                           J. Abley
Request for Comments: 4116                                           ISC
Category: Informational                                     K. Lindqvist
                                               Netnod Internet Exchange
                                                              E. Davies
                                                 Independent Researcher
                                                               B. Black
                                                        Layer8 Networks
                                                                V. Gill
                                                                    AOL
                                                              July 2005


              IPv4 Multihoming Practices and Limitations

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 (2005).

Abstract

  Multihoming is an essential component of service for many Internet
  sites.  This document describes some implementation strategies for
  multihoming with IPv4 and enumerates features for comparison with
  other multihoming proposals (particularly those related to IPv6).




















Abley, et al.                Informational                      [Page 1]

RFC 4116                    IPv4 Multihoming                   July 2005


Table of Contents

  1. Introduction ....................................................3
  2. Terminology .....................................................3
  3. IPv4 Multihoming Practices ......................................4
     3.1. Multihoming with BGP .......................................4
          3.1.1. Addressing Considerations ...........................4
          3.1.2. AS Number Considerations ............................6
     3.2. Multiple Attachments to a Single Transit Provider ..........6
     3.3. NAT- or RFC2260-based Multihoming ..........................7
  4. Features of IPv4 Multihoming ....................................7
     4.1. Redundancy .................................................7
     4.2. Load Sharing ...............................................8
     4.3. Performance ................................................8
     4.4. Policy .....................................................8
     4.5. Simplicity .................................................9
     4.6. Transport-Layer Survivability ..............................9
     4.7. Impact on DNS ..............................................9
     4.8. Packet Filtering ...........................................9
     4.9. Scalability ................................................9
     4.10. Impact on Routers ........................................10
     4.11. Impact on Hosts ..........................................10
     4.12. Interactions between Hosts and the Routing System ........10
     4.13. Operations and Management ................................10
     4.14. Cooperation between Transit Providers ....................10
  5. Security Considerations ........................................10
  6. Acknowledgements ...............................................10
  7. Informative References .........................................11























Abley, et al.                Informational                      [Page 2]

RFC 4116                    IPv4 Multihoming                   July 2005


1.  Introduction

  Multihoming is an important component of service for many Internet
  sites.  Current IPv4 multihoming practices have been added on to the
  Classless Inter Domain Routing (CIDR) architecture [RFC1519], which
  assumes that routing table entries can be aggregated based upon a
  hierarchy of customers and service providers.

  Multihoming is a mechanism by which sites can satisfy a number of
  high-level requirements.  It is widely used in the IPv4 Internet.
  There are some practical limitations, however, including concerns as
  to how it would scale with future Internet growth.  This document
  aims to document common IPv4 multihoming practices and enumerate
  their features for comparison with other multihoming approaches.

  There are a number of different ways to route and manage traffic in
  and out of a multihomed site: the majority rely on the routing policy
  capabilities of the inter-domain routing protocol, the Border Gateway
  Protocol, version 4 (BGP) [RFC1771].  This document also discusses a
  multi-homing strategy which does not rely on the capabilities of BGP.

2.  Terminology

  A "site" is an entity autonomously operating a network using IP, and
  in particular, determining the addressing plan and routing policy for
  that network.  This definition is intended to be equivalent to
  'enterprise' as defined in [RFC1918].

  A "transit provider" operates a site that directly provides
  connectivity to the Internet to one or more external sites.  The
  connectivity provided extends beyond the transit provider's own site
  and its own direct customer networks.  A transit provider's site is
  directly connected to the sites for which it provides transit.

  A "multihomed" site is one with more than one transit provider.
  "Site-multihoming" is the practice of arranging a site to be
  multihomed.

  The term "re-homing" denotes a transition of a site between two
  states of connectedness, due to a change in the connectivity between
  the site and its transit providers' sites.

  A "multi-attached" site has more than one point of layer-3
  interconnection to a single transit provider.

  Provider-Independent (PI) addresses are globally-unique addresses
  which are not assigned by a transit provider, but are provided by
  some other organisation, usually a Regional Internet Registry (RIR).



Abley, et al.                Informational                      [Page 3]

RFC 4116                    IPv4 Multihoming                   July 2005


  Provider-Aggregatable (PA) addresses are globally-unique addresses
  assigned by a transit provider to a customer.  The addresses are
  considered "aggregatable" because the set of routes corresponding to
  the PA addresses are usually covered by an aggregate route set
  corresponding to the address space operated by the transit provider,
  from which the assignment was made.

  Note that the words "assign" and "allocate" have specific meanings in
  Regional Internet Registry (RIR) address management policies, but are
  used more loosely in this document.

3.  IPv4 Multihoming Practices

3.1.  Multihoming with BGP

  The general approach for multihoming with BGP is to announce a set of
  routes to two or more transit providers.  This provides the rest of
  the Internet with multiple paths back to the multihomed sites, and
  each transit provider provides an additional possible path for the
  site's outbound traffic.

3.1.1.  Addressing Considerations

3.1.1.1.  PI Addresses

  The site uses PI addresses, and a set of routes covering those PI
  addresses is announced or propagated by two or more transit
  providers.

  Using PI addresses has long been the preferred approach for IPv4
  multihoming.  Until the mid-1990s this was relatively easy to
  accomplish, as the maximum generally accepted prefix length in the
  global routing table was a /24, and little justification was needed
  to obtain a /24 PI assignment.  Since then, RIR address management
  policies have become less liberal in this respect.  Not all RIRs
  support the assignment of address blocks to small, multihomed end-
  users, and those that do support it require justification for blocks
  as large as a /24, which cannot be met by small sites.  As a
  consequence, PI addresses are not available to many sites who wish to
  multihome.

  Each site that uses PI addresses introduces an additional prefix into
  the global routing system.  If this scheme for multihoming became
  widespread, it would present scaling concerns.







Abley, et al.                Informational                      [Page 4]

RFC 4116                    IPv4 Multihoming                   July 2005


3.1.1.2.  PA Addresses

  The site uses PA addresses assigned by a single transit provider.
  The set of routes covering those PA addresses (the "site route set")
  is announced or propagated by one or more additional transit
  providers.  The transit provider which assigned the PA addresses (the
  "primary transit provider") originates a set of routes which cover
  the site route set.  The primary transit provider often originates or
  propagates the site route set as well as the covering aggregates.

  The use of PA addresses is applicable to sites whose addressing
  requirements are not sufficient to meet the requirements for PI
  assignments by RIRs.  However, in the case where the site route set
  is to be announced or propagated by two or more different transit
  providers, common operational practice still dictates minimum /24
  prefixes, which may be larger than the allocation available to small
  sites.

  There have been well-documented examples of sites filtering long-
  prefix routes which are covered by a transit-providers aggregate.  If
  this practice were to become very widespread, it might limit the
  effectiveness of multihoming using PA addresses.  However, limited
  filtering of this kind can be tolerated because the aggregate
  announcements of the primary transit provider should be sufficient to
  attract traffic from autonomous systems which do not accept the
  covered site route set.  The more traffic that follows the primary
  transit provider's aggregate in the absence of the covered, more-
  specific route, the greater the reliance on that primary transit
  provider.  In some cases, this reliance might result in an effective
  single point of failure.

  Traffic following the primary transit provider's aggregate routes may
  still be able to reach the multihomed site, even in the case where
  the connection between the primary transit provider and the site has
  failed.  The site route set will still be propagating through the
  site's other transit providers.  If that route set reaches (and is
  accepted by) the primary transit provider, connectivity for traffic
  following the aggregate route will be preserved.

  Sites that use PA addresses are usually obliged to renumber if they
  decide not to retain connectivity to the primary transit provider.
  While this is a common requirement for all sites using PA addresses
  (and not just those that are multihomed), it is one that may have
  more frequent impact on sites whose motivation to multihome is to
  facilitate changes of ISP.  A multihomed site using PA addresses can
  still add or drop other service providers without having to renumber.





Abley, et al.                Informational                      [Page 5]

RFC 4116                    IPv4 Multihoming                   July 2005


3.1.2.  AS Number Considerations

3.1.2.1.  Consistent Origin AS

  A multihomed site may choose to announce routes to two or more
  transit providers from a globally-unique Autonomous System (AS)
  number assigned to the site.  This causes the origin of the route to
  appear consistent when viewed from all parts of the Internet.

3.1.2.2.  Inconsistent Origin AS

  A multihomed site may choose to use a private-use AS number [RFC1930]
  to originate routes to transit providers.  It is normal practice for
  private-use AS numbers to be stripped from AS_PATH attributes before
  they are allowed to propagate from transit providers towards peers.
  Therefore, routes observed from other parts of the Internet may
  appear to have inconsistent origins.

  When using private-use AS numbers, collisions between the use of
  individual numbers by different transit providers are possible.
  These collisions are arguably best avoided by not using private-use
  AS numbers for applications which involve routing across
  administrative domain boundaries.

  A multihomed site may request that their transit providers each
  originate the site's routes from the transit providers' ASes.
  Dynamic routing (for the purposes of withdrawing the site's route in
  the event that connectivity to the site is lost) is still possible,
  in this case, using the transit providers' internal routing systems
  to trigger the externally-visible announcements.

  Operational troubleshooting is facilitated by the use of a consistent
  origin AS.  This allows import policies to be based on a route's true
  origin rather than on intermediate routing details, which may change
  (e.g., as transit providers are added and dropped by the multihomed
  site).

3.2.  Multiple Attachments to a Single Transit Provider

  Multihoming can be achieved through multiple connections to a single
  transit provider.  This imposes no additional load on the global
  routing table beyond that involved in the site being single-attached.
  A site that has solved its multihoming needs in this way is commonly
  referred to as "multi-attached".







Abley, et al.                Informational                      [Page 6]

RFC 4116                    IPv4 Multihoming                   July 2005


  It is not a requirement that the multi-attached site exchange routing
  information with its transit provider using BGP.  However, in the
  event of failure, some mechanism for re-routing inbound and outbound
  traffic over remaining circuits is required.  BGP is often used for
  this purpose.

  Multi-attached sites gain no advantages from using PI addresses or
  (where BGP is used) globally-unique AS numbers, and have no need to
  be able to justify address assignments of a particular minimum size.
  However, multi-attachment does not protect a site from the failure of
  the single transit provider.

3.3.  NAT- or RFC2260-based Multihoming

  This method uses PA addresses assigned by each transit provider to
  which the site is connected.  The addresses are either allocated to
  individual hosts within the network according to [RFC2260], or the
  site uses Network Address Translation (NAT) to translate the various
  provider addresses into a single set of private-use addresses
  [RFC1918] within the site.  The site is effectively singlehomed to
  more than one transit provider.  None of the transit providers need
  to make any accommodations beyond those typically made for a non-
  multihomed customer.

  This approach accommodates a wide range of sites, from residential
  Internet users to very large enterprises, requires no PI addresses or
  AS numbers, and imposes no additional load on the Internet's global
  routing system.  However, it does not address several common
  motivations for multihoming, most notably transport-layer
  survivability.

4.  Features of IPv4 Multihoming

  The following sections describe some of the features of the
  approaches described in Section 3, in the context of the general
  goals for multihoming architectures presented in [RFC3582].  Detailed
  descriptions and rationale for these goals can be found in that
  document.

4.1.  Redundancy

  All the methods described provide redundancy, which can protect a
  site from some single-point failures.  The degree of protection
  depends on the choice of transit providers and the methods used to
  interconnect the site to those transit providers.






Abley, et al.                Informational                      [Page 7]

RFC 4116                    IPv4 Multihoming                   July 2005


4.2.  Load Sharing

  All of the methods described provide some measure of load-sharing
  capability.  Outbound traffic can be shared across ISPs using
  appropriate exit selection policies; inbound traffic can be
  distributed using appropriate export policies designed to influence
  the exit selection of remote sites sending traffic back towards the
  multihomed site.

  In the case of RFC2260/NAT multihoming, distribution of inbound
  traffic is controlled by address selection on the host or NAT.

4.3.  Performance

  BGP-speaking sites can employ import policies that cause exit
  selection to avoid paths known to be problematic.  For inbound
  traffic, sites can often employ route export policy, which affords
  different treatment of traffic towards particular address ranges
  within their network.

  It should be noted that this is not a comprehensive capability.  In
  general, there are many traffic engineering goals which can only be
  loosely approximated using this approach.

  In the case of RFC2260/NAT multihoming in the absence of BGP routing
  information, management of outbound traffic is not possible.  The
  path taken by inbound traffic for a particular session can be
  controlled by source address selection on the host or NAT.

4.4.  Policy

  In some circumstances, it is possible to route traffic of a
  particular type (e.g., protocol) via particular transit providers.
  This can be done if the devices in the site which source or sink that
  traffic can be isolated to a set of addresses to which a special
  export policy can be applied.

  An example of this capability is the grouping of budget, best-effort
  Internet customers into a particular range of addresses that is
  covered by a route which is announced preferentially over a single,
  low-quality transit path.

  In the case of RFC2260/NAT multihoming, policies such as those
  described here can be accommodated by appropriate address selection
  on the host or NAT.  More flexible implementations may be possible
  for sessions originated from the multihomed site by selecting an
  appropriate source address on a host or NAT, according to criteria
  such as transport-layer protocols and addresses (ports).



Abley, et al.                Informational                      [Page 8]

RFC 4116                    IPv4 Multihoming                   July 2005


4.5.  Simplicity

  The current methods used as multihoming solutions are not without
  their complexities, but have proven to be sufficiently simple to be
  used.  They have the advantage of familiarity due to having been
  deployed extensively.

4.6.  Transport-Layer Survivability

  All BGP-based multihoming practices provide some degree of session
  survivability for transport-layer protocols.  However, in cases where
  path convergence takes a long time following a re-homing event,
  sessions may time out.

  Transport-layer sessions will not, in general, survive over a re-
  homing event when using RFC2260/NAT multihoming.  Transport protocols
  which support multiple volatile endpoint addresses may be able to
  provide session stability; however, these transport protocols are not
  in wide use.

  In all the methods described in this document, new transport-layer
  sessions are able to be created following a re-homing event.

4.7.  Impact on DNS

  These multihoming strategies impose no new requirements on the DNS.

4.8.  Packet Filtering

  These multihoming practices do not preclude filtering of packets with
  inappropriate source or destination addresses at the administrative
  boundary of the multihomed site.

4.9.  Scalability

  Current IPv4 multihoming practices are thought to contribute to
  significant observed growth in the amount of state held in the global
  inter-provider routing system.  This is a concern because of both the
  hardware requirements it imposes and the impact on the stability of
  the routing system.  This issue is discussed in greater detail in
  [RFC3221].

  Of the methods presented in this document, RFC2260/NAT multihoming
  and multi-attaching to a single transit provider provide no
  additional state to be held in the global routing system.  All other
  strategies contribute to routing system state bloat.





Abley, et al.                Informational                      [Page 9]

RFC 4116                    IPv4 Multihoming                   July 2005


  Globally-unique AS numbers are a finite resource.  Thus, widespread
  multihoming that uses strategies requiring assignment of AS numbers
  might lead to increased resource contention.

4.10.  Impact on Routers

  For some of the multihoming approaches described in this document,
  the routers at the boundary of the multihomed site are required to
  participate in BGP sessions with transit provider routers.  Other
  routers within the site generally have no special requirements beyond
  those in singlehomed sites.

4.11.  Impact on Hosts

  There are no requirements of hosts beyond those in singlehomed sites.

4.12.  Interactions between Hosts and the Routing System

  There are no requirements for interaction between routers and hosts
  beyond those in singlehomed sites.

4.13.  Operations and Management

  There is extensive operational experience in managing IPv4-multihomed
  sites.

4.14.  Cooperation between Transit Providers

  Transit providers who are asked to announce or propagate a PA prefix
  covered by some other (primary) transit provider usually obtain
  authorisation first.  However, there is no technical requirement or
  common contractual policy which requires this coordination to take
  place.

5.  Security Considerations

  This document discusses current IPv4 multihoming practices, but
  provides no analysis of the security implications of multihoming.

6.  Acknowledgements

  Special acknowledgement goes to John Loughney for proof-reading and
  corrections.  Thanks also goes to Pekka Savola and Iljitsch van
  Beijnum for providing feedback and contributing text.

  This work was supported by the US National Science Foundation
  (research grant SCI-0427144) and DNS-OARC.




Abley, et al.                Informational                     [Page 10]

RFC 4116                    IPv4 Multihoming                   July 2005


7.  Informative References

  [RFC1519]  Fuller, V., Li, T., Yu, J., and K. Varadhan, "Classless
             Inter-Domain Routing (CIDR): an Address Assignment and
             Aggregation Strategy", RFC 1519, September 1993.

  [RFC1771]  Rekhter, Y. and T. Li, "A Border Gateway Protocol 4
             (BGP-4)", RFC 1771, March 1995.

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

  [RFC1930]  Hawkinson, J. and T. Bates, "Guidelines for creation,
             selection, and registration of an Autonomous System (AS)",
             BCP 6, RFC 1930, March 1996.

  [RFC2260]  Bates, T. and Y. Rekhter, "Scalable Support for Multi-
             homed Multi-provider Connectivity", RFC 2260,
             January 1998.

  [RFC3221]  Huston, G., "Commentary on Inter-Domain Routing in the
             Internet", RFC 3221, December 2001.

  [RFC3582]  Abley, J., Black, B., and V. Gill, "Goals for IPv6 Site-
             Multihoming Architectures", RFC 3582, August 2003.

























Abley, et al.                Informational                     [Page 11]

RFC 4116                    IPv4 Multihoming                   July 2005


Authors' Addresses

  Joe Abley
  Internet Systems Consortium, Inc.
  950 Charter Street
  Redwood City, CA  94063
  USA

  Phone: +1 650 423 1317
  EMail: [email protected]


  Kurt Erik Lindqvist
  Netnod Internet Exchange
  Bellmansgatan 30
  Stockholm  S-118 47
  Sweden

  Phone: +46 8 615 85 70
  EMail: [email protected]


  Elwyn B. Davies
  Independent Researcher
  Soham, Cambridgeshire  CB7 5AW
  UK

  Phone: +44 7889 488 335
  EMail: [email protected]


  Benjamin Black
  Layer8 Networks

  EMail: [email protected]


  Vijay Gill
  AOL
  12100 Sunrise Valley Dr
  Reston, VA  20191
  US

  Phone: +1 410 336 4796
  EMail: [email protected]






Abley, et al.                Informational                     [Page 12]

RFC 4116                    IPv4 Multihoming                   July 2005


Full Copyright Statement

  Copyright (C) The Internet Society (2005).

  This document is subject to the rights, licenses and restrictions
  contained in BCP 78, and except as set forth therein, the authors
  retain all their rights.

  This document and the information contained herein are provided on an
  "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
  OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
  ENGINEERING TASK FORCE DISCLAIM 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.

Intellectual Property

  The IETF takes no position regarding the validity or scope of any
  Intellectual Property Rights or other rights that might be claimed to
  pertain to the implementation or use of the technology described in
  this document or the extent to which any license under such rights
  might or might not be available; nor does it represent that it has
  made any independent effort to identify any such rights.  Information
  on the procedures with respect to rights in RFC documents can be
  found in BCP 78 and BCP 79.

  Copies of IPR disclosures made to the IETF Secretariat and any
  assurances of licenses to be made available, or the result of an
  attempt made to obtain a general license or permission for the use of
  such proprietary rights by implementers or users of this
  specification can be obtained from the IETF on-line IPR repository at
  http://www.ietf.org/ipr.

  The IETF invites any interested party to bring to its attention any
  copyrights, patents or patent applications, or other proprietary
  rights that may cover technology that may be required to implement
  this standard.  Please address the information to the IETF at ietf-
  [email protected].

Acknowledgement

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







Abley, et al.                Informational                     [Page 13]