MPLS Working Group Santosh Esale
INTERNET-DRAFT Raveendra Torvi
Intended Status: Proposed Standard Juniper Networks
Expires: November 20, 2016
Luyuan Fang
Microsoft
Luay Jalil
Verizon
May 19, 2016
Fast Reroute for Node Protection in LDP-based LSPs
draft-esale-mpls-ldp-node-frr-03
Abstract
This document describes procedures to support node protection for
unicast Label Switched Paths (LSPs) established by Label Distribution
Protocol (LDP). In order to protect a node N, the Point of Local
Repair (PLR) of N must discover the Merge Points (MPs) of node N such
that traffic can be redirected to them in case of node N failure.
Redirecting the traffic around the failed node N depends on existing
point-to-point LSPs originated from the PLR to the MPs while
bypassing the protected node N. The procedures described in this
document are topology independent in a sense that they provide node
protection in any topology.
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance with the
provisions of BCP 78 and BCP 79.
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
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/1id-abstracts.html
Esale, et al. Expires November 20, 2016 [Page 1]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
Copyright and License 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. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Merge Point (MP) Discovery . . . . . . . . . . . . . . . . . . 3
4. Constructing Bypass LSPs . . . . . . . . . . . . . . . . . . . 4
5. Obtaining Label Mapping from MP . . . . . . . . . . . . . . . . 5
6. Forwarding Considerations . . . . . . . . . . . . . . . . . . . 6
7. Synergy with node protection in mLDP . . . . . . . . . . . . . 6
8. Security Considerations . . . . . . . . . . . . . . . . . . . . 6
9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . . 6
10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 6
11. Normative References . . . . . . . . . . . . . . . . . . . . . 6
12. Informative References . . . . . . . . . . . . . . . . . . . . 7
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 7
Esale, et al. Expires November 20, 2016 [Page 2]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
1. Terminology
PLR: Point of Local Repair (the LSR that redirects the traffic to
one or more Merge Point LSRs).
MP: Merge Point. Any LSR on the LDP-signaled (multi-point to
point) LSP, provided that the path from that LSR to the
egress of that LSP is not affected by the failure of the
protected node.
tLDP: A targeted LDP session is an LDP session between non-directly
connected LSRs, established using the LDP extended discovery
mechanism.
FEC: Forwarding equivalence class.
IGP: Interior Gateway Protocol.
BR: Border Router.
2. Introduction
This document describes procedures to support node protection for
unicast Label Switched Paths (LSPs) established by Label Distribution
Protocol (LDP) [RFC5036]. In order to protect a node N, the Point of
Local Repair (PLR) of N must discover the Merge Points (MPs) of node
N such that traffic can be redirected to them in case of node N
failure. Redirecting the traffic around the failed node N depends on
existing explicit path Point-to-Point (P2P) LSPs originated from the
PLR LSR to the MPs while bypassing node N. The procedures to setup
these P2P LSPs are outside the scope of this document, but one option
is to use RSVP-TE based techniques [RFC3209] to accomplish it.
Finally, sending traffic from the PLR to the MPs requires the PLR to
obtain FEC-label bindings from the MPs. The procedures described in
this document relies on Targeted LDP (tLDP) session [RFC5036] for the
PLR to obtain such FEC-Label bindings. The procedure described in
this document assumes the use of platform-wide label space. The
procedures for node protection described in this document fall into
the category of local protection. The procedures described in this
document apply to LDP LSPs bound to either an IPv4 or IPv6 Prefix FEC
element. The procedures described in this document are topology
independent in a sense that they provide node protection in any
topology. Thus these procedures provide topology independent fast
reroute.
3. Merge Point (MP) Discovery
Esale, et al. Expires November 20, 2016 [Page 3]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
For a given LSP that traverses the PLR, the protected node N, and a
particular neighbor of the protected node, we'll refer to this
neighbor as the "next next-hop". Note that from the PLR's perspective
the protected node N is the next hop for the FEC associated with that
LSP. Likewise, from the protected node's perspective the next next-
hop is the next hop for that FEC. If for a given <LSP, PLR, N>
triplet the next next-hop is in the same routing subdomain (area) as
the PLR, then that next next-hop acts as the MP for that triplet. For
a given LSP traversing a PLR and the node protected by the PLR, the
PLR discovers its next next-hops (MPs) that are in the same routing
subdomain (IGP area) as the PLR from IGP shortest path first (SPF)
calculations. The discovery of next next-hop, depending on an
implementation, may not involve any additional SPF, above and beyond
what will be needed by ISIS/OSPF anyway, as the next next-hop, just
like the next-hop, is a by-product of SPF computation. Also, the PLR
may discover all possible MPs from either its traffic engineering
database or link state database. Some implementations MAY need
appropriate configuration to populate the traffic engineering
database. The traffic engineering database is populated by routing
protocols such as ISIS and OSPF or configured statically. If for a
given <LSP, PLR, N> triplet the node protected by the PLR is an
Border Router (BR), then the PLR and the next next-hop may end up in
different routing subdomain. This could happen when an LSP traversing
the PLR and the protected node does not terminate in the same routing
subdomain as the PLR. In this situation the PLR may not be able to
determine the next next-hop from shortest path first (SPF)
calculations, and thus may not be able to use the next next-hop as
the MP. In this scenario the PLR uses an "alternative" BR as the MP,
where an alternative BR is defined as follows. For a given LSP that
traverses the PLR and the (protected) BR, an alternative BR is
defined as any BR that advertises into PLR's own routing subdomain
reachability to the FEC associated with the LSP. Note that even if a
PLR protects an BR, for some of the LSPs traversing the PLR and the
BR, the next next-hops may be in the same routing subdomain as the
PLR, in which case these next next-hops act as MPs for these LSPs.
Note that even if the protected node is not an BR, if an LSP
traversing the PLR and the protected node does not terminate in the
same routing subdomain as the PLR, then for this LSP the PLR MAY use
an alternative BR (as defined above), rather than the next next-hop
as the MP. When there are several candidate BRs for alternative BR,
the LSR MUST select one BR. The algorithm used for the alternative BR
selection is a local matter but one option is to select the BR per
FEC based on shortest path from PLR to the BR.
4. Constructing Bypass LSPs
Esale, et al. Expires November 20, 2016 [Page 4]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
As mentioned before, redirecting traffic around the failed node N
depends on existing explicit path Point-to-Point (P2P) LSPs
originated from the PLR to the MPs while bypassing node N. Let's
refer to these LSPs as "bypass LSPs". While the procedures to signal
these bypass LSPs are outside the scope of this document, this
document assumes use of RSVP-TE LSPs [RFC3209] to accomplish this.
Once a PLR that protects a given node N discovers the set of MPs
associated with itself and the protected node, at the minimum the PLR
MUST (automatically) establish bypass LSPs to all these MPs. The
bypass LSPs MUST be established before the failure of the protected
node. One could observe that if the protected node is not an BR and
the PLR does not use alternative BR(s) as MP(s), then the set of all
the IGP neighbors of the protected node forms a superset of the MPs.
Thus it would be sufficient for the PLR to establish bypass LSPs with
all the IGP neighbors of the protected node, even though some of
these neighbors may not be MPs for any of the LSPs traversing the PLR
and the protected node. The bypass LSPs MUST avoid traversing the
protected node, which means that the bypass LSPs are explicitly
routed LSPs. of course, using RSVP-TE to establish bypass LSPs allows
these LSPs to be explicitly routed. As a given router may act as an
MP for more than one LSP traversing the PLR, the protected node, and
the MP, the same bypass LSP will be used to protect all these LSPs.
5. Obtaining Label Mapping from MP
As mentioned before, sending traffic from the PLR to the MPs requires
the PLR to obtain FEC-label bindings from the MPs. The solution
described in this document relies on Targeted LDP (tLDP) session
[RFC5036] for the PLR to obtain such mappings. Specifically, for a
given PLR and the node protected by this PLR, at the minimum the PLR
MUST (automatically) establish tLDP with all the MPs associated with
this PLR and the protected node. These tLDP sessions MUST be
established before the failure of the protected node. One could
observe that if the protected node is not an BR and the PLR does not
use alternative BR(s) as MP(s), then the set of all the IGP neighbors
of the protected node forms a superset of the MPs. Thus it will be
sufficient for the PLR to (automatically) establish tLDP session with
all the IGP neighbors of the protected node (except the PLR) that are
in the same area as the PLR, even though some of these neighbors may
not be MPs for any of the LSPs traversing the PLR and the protected
node. At the minimum for a given tLDP peer the PLR MUST obtain FEC-
label mapping for the FEC(s) for which the peer acts as an MP. The
PLR MUST obtain this mapping before the failure of the protected
node. To obtain this mapping for only these FECs and no other FECs
that the peer may maintain, the PLR SHOULD rely on the LDP Downstream
on Demand (DoD) procedures [RFC5036]. Otherwise, without relying on
the DoD procedures, the PLR may end up receiving from a given tLDP
peer FEC-label mappings for all the FECs maintained by the peer, even
Esale, et al. Expires November 20, 2016 [Page 5]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
if the peer does not act as an MP for some of these FECs. If the LDP
DoD procedures are not used, then for the purpose of the procedures
specified in this draft the only label mappings that SHOULD be
exchanged are for the Prefix FEC elements whose PreLen value is
either 32 (IPv4), or 128 (IPv6); label mappings for the Prefix FEC
elements with any other PreLen value SHOULD NOT be exchanged. When a
PLR has one or more BRs acting as MPs, the PLR MAY use the procedures
specified in [draft-ietf-mpls-app-aware-tldp] to limit the set of
FEC-label mappings received from non-BR MPs to only the mappings for
the FECs associated with the LSPs that terminate in the PLR's own
routing subdomain (area).
6. Forwarding Considerations
When a PLR detects failure of the protected node then rather than
swapping an incoming label with a label that the PLR received from
the protected node, the PLR swaps the incoming label with the label
that the PLR receives from the MP, and then pushes the label
associated with the bypass LSP to that MP. To minimize micro-loop
during the IGP global convergence PLR may continue to use the bypass
LSP during network convergence by adding small delay before switching
to a new path.
7. Synergy with node protection in mLDP
Both the bypass LSPs and tLDP sessions described in this document
could also be used for the purpose of mLDP node protection, as
described in [draft-ietf-mpls-mldp-node-protection].
8. Security Considerations
The same security considerations apply as those for the base LDP
specification, as described in [RFC5036].
9. IANA Considerations
This document introduces no new IANA Considerations.
10. Acknowledgements
We are indebted to Yakov Rekhter for many discussions on this topic.
We like to thank Hannes Gredler, Aman Kapoor, Minto Jeyananth, Eric
Rosen and Vladimir Blazhkun for through review of this document.
11. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
Esale, et al. Expires November 20, 2016 [Page 6]
INTERNET DRAFT <draft-esale-mpls-ldp-node-frr> May 19, 2016
[RFC3209] D. Awduche, et al., "RSVP-TE: Extensions to RSVP for LSP
Tunnels", RFC3209, Decembet 2001
[RFC5036] Andersson, L., Minei, I., and B. Thomas, "LDP
Specification", RFC 5036, October 2007.
[draft-ietf-mpls-app-aware-tldp] Esale, S., et al.,"Application-
aware Targeted LDP", draft-esale-mpls-app-aware-tldp, work
in progress
12. Informative References
[draft-ietf-mpls-mldp-node-protection], IJ. Wijnands, et al., "mLDP
Node Protection", draft-ietf-mpls-mldp-node-protection,
work in progress
Authors' Addresses
Santosh Esale
Juniper Networks
EMail: sesale@juniper.net
Raveendra Torvi
Juniper Networks
EMail: rtorvi@juniper.net
Luyuan Fang
Microsoft
Email: lufang@microsoft.com
Luay Jalil
Verizon
Email: luay.jalil@verizon.com
Esale, et al. Expires November 20, 2016 [Page 7]