BGP ADD-PATH: Advertisement of Multiple Paths in BGP
How BGP speakers can advertise several paths to the same prefix without each one replacing the last.
Walton, D., Retana, A., Chen, E., and J. Scudder, "Advertisement of Multiple Paths in BGP", RFC 7911, DOI 10.17487/RFC7911, July 2016, <https://www.rfc-editor.org/info/rfc7911>.
@misc{rfc7911,
series = {Request for Comments},
number = 7911,
howpublished = {RFC 7911},
publisher = {RFC Editor},
doi = {10.17487/RFC7911},
url = {https://www.rfc-editor.org/info/rfc7911},
author = {D. Walton and A. Retana and E. Chen and J. Scudder},
title = {{Advertisement of Multiple Paths in BGP}},
pagetotal = 8,
year = 2016,
month = jul,
} RFC text
Contents
Internet Engineering Task Force (IETF) D. Walton
Request for Comments: 7911 Cumulus Networks
Category: Standards Track A. Retana
ISSN: 2070-1721 E. Chen
Cisco Systems, Inc.
J. Scudder
Juniper Networks
July 2016
Advertisement of Multiple Paths in BGP
Abstract
This document defines a BGP extension that allows the advertisement
of multiple paths for the same address prefix without the new paths
implicitly replacing any previous ones. The essence of the extension
is that each path is identified by a Path Identifier in addition to
the address prefix.
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 7841.
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/rfc7911.
Copyright Notice
Copyright (c) 2016 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1. Specification of Requirements . . . . . . . . . . . . . . 2
2. How to Identify a Path . . . . . . . . . . . . . . . . . . . 3
3. Extended NLRI Encodings . . . . . . . . . . . . . . . . . . . 3
4. ADD-PATH Capability . . . . . . . . . . . . . . . . . . . . . 4
5. Operation . . . . . . . . . . . . . . . . . . . . . . . . . . 4
6. Deployment Considerations . . . . . . . . . . . . . . . . . . 5
7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 6
8. Security Considerations . . . . . . . . . . . . . . . . . . . 6
9. References . . . . . . . . . . . . . . . . . . . . . . . . . 6
9.1. Normative References . . . . . . . . . . . . . . . . . . 6
9.2. Informative References . . . . . . . . . . . . . . . . . 7
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . 8
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 8
1. Introduction
The BGP specification [RFC4271] defines an Update-Send Process to
advertise the routes chosen by the Decision Process to other BGP
speakers. No provisions are made to allow the advertisement of
multiple paths for the same address prefix or Network Layer
Reachability Information (NLRI). In fact, a route with the same NLRI
as a previously advertised route implicitly replaces the previous
advertisement.
This document defines a BGP extension that allows the advertisement
of multiple paths for the same address prefix without the new paths
implicitly replacing any previous ones. The essence of the extension
is that each path is identified by a Path Identifier in addition to
the address prefix.
The availability of the additional paths can help reduce or eliminate
persistent route oscillations [RFC3345]. It can also help with
optimal routing and routing convergence in a network by providing
potential alternate or backup paths, respectively.
1.1. Specification of Requirements
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. How to Identify a Path
As defined in [RFC4271], a path refers to the information reported in
the Path Attribute field of an UPDATE message. As the procedures
specified in [RFC4271] allow only the advertisement of one path for a
particular address prefix, a path for an address prefix from a BGP
peer can be keyed on the address prefix.
In order for a BGP speaker to advertise multiple paths for the same
address prefix, a new identifier (termed "Path Identifier" hereafter)
needs to be introduced so that a particular path for an address
prefix can be identified by the combination of the address prefix and
the Path Identifier.
The assignment of the Path Identifier for a path by a BGP speaker is
purely a local matter. However, the Path Identifier MUST be assigned
in such a way that the BGP speaker is able to use the (Prefix, Path
Identifier) to uniquely identify a path advertised to a neighbor. A
BGP speaker that re-advertises a route MUST generate its own Path
Identifier to be associated with the re-advertised route. A BGP
speaker that receives a route should not assume that the identifier
carries any particular semantics.
3. Extended NLRI Encodings
In order to carry the Path Identifier in an UPDATE message, the NLRI
encoding MUST be extended by prepending the Path Identifier field,
which is of four octets.
For example, the NLRI encoding specified in [RFC4271] is extended as
the following:
+--------------------------------+
| Path Identifier (4 octets) |
+--------------------------------+
| Length (1 octet) |
+--------------------------------+
| Prefix (variable) |
+--------------------------------+
The usage of the extended NLRI encodings is specified in Section 5.
4. ADD-PATH Capability
The ADD-PATH Capability is a BGP capability [RFC5492], with
Capability Code 69. The Capability Length field of this capability
is variable. The Capability Value field consists of one or more of
the following tuples:
+------------------------------------------------+
| Address Family Identifier (2 octets) |
+------------------------------------------------+
| Subsequent Address Family Identifier (1 octet) |
+------------------------------------------------+
| Send/Receive (1 octet) |
+------------------------------------------------+
The meaning and use of the fields are as follows:
Address Family Identifier (AFI):
This field is the same as the one used in [RFC4760].
Subsequent Address Family Identifier (SAFI):
This field is the same as the one used in [RFC4760].
Send/Receive:
This field indicates whether the sender is (a) able to receive
multiple paths from its peer (value 1), (b) able to send
multiple paths to its peer (value 2), or (c) both (value 3) for
the <AFI, SAFI>.
If any other value is received, then the capability SHOULD be
treated as not understood and ignored [RFC5492].
A BGP speaker that wishes to indicate support for multiple AFI/SAFIs
MUST do so by including the information in a single instance of the
ADD-PATH Capability.
5. Operation
The Path Identifier specified in Section 3 can be used to advertise
multiple paths for the same address prefix without subsequent
advertisements replacing the previous ones. Apart from the fact that
this is now possible, the route advertisement rules of [RFC4271] are
not changed. In particular, a new advertisement for a given address
prefix and a given Path Identifier replaces a previous advertisement
for the same address prefix and Path Identifier. If a BGP speaker
receives a message to withdraw a prefix with a Path Identifier not
seen before, it SHOULD silently ignore it.
For a BGP speaker to be able to send multiple paths to its peer, that
BGP speaker MUST advertise the ADD-PATH Capability with the Send/
Receive field set to either 2 or 3, and MUST receive from its peer
the ADD-PATH Capability with the Send/Receive field set to either 1
or 3, for the corresponding <AFI, SAFI>.
A BGP speaker MUST follow the procedures defined in [RFC4271] when
generating an UPDATE message for a particular <AFI, SAFI> to a peer
unless the BGP speaker advertises the ADD-PATH Capability to the peer
indicating its ability to send multiple paths for the <AFI, SAFI>,
and also receives the ADD-PATH Capability from the peer indicating
its ability to receive multiple paths for the <AFI, SAFI>, in which
case the speaker MUST generate a route update for the <AFI, SAFI>
based on the combination of the address prefix and the Path
Identifier, and use the extended NLRI encodings specified in this
document. The peer SHALL act accordingly in processing an UPDATE
message related to a particular <AFI, SAFI>.
A BGP speaker SHOULD include the best route [RFC4271] when more than
one path is advertised to a neighbor, unless it is a path received
from that neighbor.
As the Path Identifiers are locally assigned, and may or may not be
persistent across a control plane restart of a BGP speaker, an
implementation SHOULD take special care so that the underlying
forwarding plane of a "Receiving Speaker" as described in [RFC4724]
is not affected during the graceful restart of a BGP session.
6. Deployment Considerations
The extension proposed in this document provides a mechanism for a
BGP speaker to advertise multiple paths over a BGP session. Care
needs to be taken in its deployment to ensure consistent routing and
forwarding in a network [ADDPATH].
The only explicit indication that the encoding described in Section 3
is in use in a particular BGP session is the exchange of Capabilities
described in Section 4. If the exchange is successful [RFC5492],
then the BGP speakers will be able to process all BGP UPDATES
properly, as described in Section 5. However, if, for example, a
packet analyzer is used on the wire to examine an active BGP session,
it may not be able to properly decode the BGP UPDATES because it
lacks prior knowledge of the exchanged Capabilities.
When deployed as a provider edge router or a peering router that
interacts with external neighbors, a BGP speaker usually advertises
at most one path to the internal neighbors in a network. In the case
where the speaker is configured to advertise multiple paths to the
internal neighbors, and additional information is needed for the
application, the speaker could use attributes such as the
Edge_Discriminator attribute [FAST]. The use of that type of
additional information is outside the scope of this document.
7. IANA Considerations
IANA has assigned the value 69 for the ADD-PATH Capability described
in this document. This registration is in the "Capability Codes"
registry.
8. Security Considerations
This document defines a BGP extension that allows the advertisement
of multiple paths for the same address prefix without the new paths
implicitly replacing any previous ones. As a result, multiple paths
for a large number of prefixes may be received by a BGP speaker,
potentially depleting memory resources or even causing network-wide
instability, which can be considered a denial-of-service attack.
Note that this is not a new vulnerability, but one that is present in
the base BGP specification [RFC4272].
The use of the ADD-PATH Capability is intended to address specific
needs related to, for example, eliminating route oscillations that
were induced by the MULTI_EXIT_DISC (MED) attribute [STOP-OSC].
While describing the applications for the ADD-PATH Capability is
outside the scope of this document, users are encouraged to examine
their behavior and potential impact by studying the best practices
described in [ADDPATH].
Security concerns in the base operation of BGP [RFC4271] also apply.
9. References
9.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119,
DOI 10.17487/RFC2119, March 1997,
<http://www.rfc-editor.org/info/rfc2119>.
[RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A
Border Gateway Protocol 4 (BGP-4)", RFC 4271,
DOI 10.17487/RFC4271, January 2006,
<http://www.rfc-editor.org/info/rfc4271>.
[RFC4760] Bates, T., Chandra, R., Katz, D., and Y. Rekhter,
"Multiprotocol Extensions for BGP-4", RFC 4760,
DOI 10.17487/RFC4760, January 2007,
<http://www.rfc-editor.org/info/rfc4760>.
[RFC5492] Scudder, J. and R. Chandra, "Capabilities Advertisement
with BGP-4", RFC 5492, DOI 10.17487/RFC5492, February
2009, <http://www.rfc-editor.org/info/rfc5492>.
9.2. Informative References
[ADDPATH] Uttaro, J., Francois, P., Patel, K., Haas, J., Simpson,
A., and R. Fragassi, "Best Practices for Advertisement of
Multiple Paths in IBGP", Work in Progress,
draft-ietf-idr-add-paths-guidelines-08, April 2016.
[FAST] Mohapatra, P., Fernando, R., Filsfils, C., and R. Raszuk,
"Fast Connectivity Restoration Using BGP Add-path", Work
in Progress, draft-pmohapat-idr-fast-conn-restore-03,
January 2013.
[RFC3345] McPherson, D., Gill, V., Walton, D., and A. Retana,
"Border Gateway Protocol (BGP) Persistent Route
Oscillation Condition", RFC 3345, DOI 10.17487/RFC3345,
August 2002, <http://www.rfc-editor.org/info/rfc3345>.
[RFC4272] Murphy, S., "BGP Security Vulnerabilities Analysis",
RFC 4272, DOI 10.17487/RFC4272, January 2006,
<http://www.rfc-editor.org/info/rfc4272>.
[RFC4724] Sangli, S., Chen, E., Fernando, R., Scudder, J., and Y.
Rekhter, "Graceful Restart Mechanism for BGP", RFC 4724,
DOI 10.17487/RFC4724, January 2007,
<http://www.rfc-editor.org/info/rfc4724>.
[STOP-OSC] Walton, D., Retana, A., Chen, E., and J. Scudder, "BGP
Persistent Route Oscillation Solutions", Work in Progress,
draft-ietf-idr-route-oscillation-stop-03, April 2016.
Acknowledgments
We would like to thank David Cook and Naiming Shen for their
contributions to the design and development of the extension.
Many people have made valuable comments and suggestions, including
Rex Fernando, Eugene Kim, Danny McPherson, Dave Meyer, Pradosh
Mohapatra, Keyur Patel, Robert Raszuk, Eric Rosen, Srihari Sangli,
Dan Tappan, Mark Turner, Jeff Haas, Jay Borkenhagen, Mach Chen, Denis
Ovsienko, Carlos Pignataro, Meral Shirazipour, and Kathleen Moriarty.
Authors' Addresses
Daniel Walton
Cumulus Networks
185 E. Dana Street
Mountain View, CA 94041
United States of America
Email: dwalton@cumulusnetworks.com
Alvaro Retana
Cisco Systems, Inc.
Kit Creek Rd.
Research Triangle Park, NC 27709
United States of America
Email: aretana@cisco.com
Enke Chen
Cisco Systems, Inc.
170 W. Tasman Dr.
San Jose, CA 95134
United States of America
Email: enkechen@cisco.com
John Scudder
Juniper Networks
1194 N. Mathilda Ave
Sunnyvale, CA 94089
United States of America
Email: jgs@juniper.net
AI ELI5
On this page
§1One Path per Prefix Is Not Enough
The BGP specification, RFC 4271, has an Update-Send Process that advertises the routes chosen by the Decision Process. It makes no provision for advertising multiple paths for the same address prefix, also called the same Network Layer Reachability Information (NLRI). A route with the same NLRI as an earlier route implicitly replaces it.
This RFC defines an extension that removes that limit. The essence is that each path is identified by a Path Identifier in addition to the address prefix.
The RFC says the extra paths can help reduce or eliminate persistent route oscillations. They can also help with optimal routing and convergence by providing alternate or backup paths.
Before, telling your neighbor a new route to a place made them forget the old one. This extension lets you tell them about many routes at once.
§2Identifying a Path
In BGP a path is the information in the Path Attribute field of an UPDATE message. Since base BGP allows one path per prefix, a path can be keyed on the prefix alone. To carry several, a path must be identified by the pair of the prefix and a Path Identifier.
Path Identifier assignment is a purely local matter, but it MUST let the speaker use the (Prefix, Path Identifier) pair to uniquely identify a path advertised to a neighbor. A speaker that re-advertises a route MUST generate its own Path Identifier for it. A receiver should not assume the identifier carries any particular meaning.
Each route gets a name tag that only the sender cares about. The tag just keeps routes to the same place from getting mixed up.
§3Extended NLRI Encoding
To carry the identifier in an UPDATE, the NLRI encoding MUST be extended by prepending a Path Identifier field of four octets. The rest of the entry is the usual one-octet length and the variable-length prefix.
The extended encoding is used only as described in the operation section, that is, when both sides have agreed to it.
It is like writing the name tag in front of the address on each envelope.
| Field | Name | Size | Value | Description |
|---|---|---|---|---|
PATH-ID | Path Identifier | 4 bytes | Four octets identifying this path for the prefix. | |
LENGTH | Length | 1 byte | One octet, as in the base NLRI encoding. | |
PREFIX | Prefix | 3 bytes | Variable length; drawn here only as a placeholder width. |
§4The ADD-PATH Capability
ADD-PATH is a BGP capability as defined in RFC 5492, with Capability Code 69. Its length is variable, and its value is one or more tuples of an Address Family Identifier (2 octets), a Subsequent Address Family Identifier (1 octet) and a Send/Receive field (1 octet).
The Send/Receive value says what the sender can do for that AFI/SAFI: 1 means it can receive multiple paths, 2 means it can send them, and 3 means both. Any other value SHOULD cause the capability to be treated as not understood and ignored. A speaker supporting several AFI/SAFIs MUST list them all in a single instance of the capability.
Before using the name tags, each side says whether it can send tagged routes, receive them, or both.
- 1Receive — Sender is able to receive multiple paths from its peer.
- 2Send — Sender is able to send multiple paths to its peer.
- 3Both — Sender can both send and receive multiple paths.
§5How Multiple Paths Are Sent
Apart from allowing multiple paths, the advertisement rules of RFC 4271 do not change. A new advertisement for a given prefix and Path Identifier replaces the earlier one for that same pair. A withdraw for a Path Identifier not seen before SHOULD be silently ignored.
To send multiple paths for an AFI/SAFI, a speaker MUST advertise the capability with Send/Receive of 2 or 3 and MUST receive it from the peer with a value of 1 or 3. Only then does it generate updates keyed on prefix and Path Identifier using the extended encoding. Otherwise it follows RFC 4271.
When more than one path is advertised to a neighbor, a speaker SHOULD include the best route, unless that path was received from that neighbor. Because identifiers are local and may not survive a control plane restart, implementations SHOULD take special care that the forwarding plane of a Receiving Speaker (see RFC 4724) is not affected during graceful restart.
Tagged routes are only used if one side can send them and the other can receive them. A new route with the same tag replaces the old one with that tag, and routes with other tags stay.
§6Deployment Considerations
The RFC warns that care is needed to keep routing and forwarding consistent in a network, and points to separate best-practice guidance.
The only explicit sign that the extended encoding is in use on a session is the capability exchange. A packet analyzer looking at a live session may therefore fail to decode the UPDATEs, since it lacks knowledge of the exchanged capabilities.
A provider edge or peering router facing external neighbors usually advertises at most one path to internal neighbors. If it advertises more and extra information is needed, it could use attributes such as the Edge_Discriminator attribute, which is outside the scope of this document.
If you peek at the mail without knowing about the name tags, it looks like gibberish. Careful setup keeps everyone's maps consistent.
§7Security and IANA
Multiple paths for many prefixes may reach a speaker, potentially depleting memory or even causing network-wide instability, which can be considered a denial-of-service attack. The RFC notes this is not a new vulnerability but one present in base BGP. Users are encouraged to study best practices before use, and the security concerns of base BGP also apply.
IANA assigned value 69 for the ADD-PATH Capability in the Capability Codes registry.
If someone sends you a mountain of routes, your router could run out of room. That risk already existed, but this extension can make the pile bigger.