MVPN Explicit Tracking: Explicit Tracking with Wildcard Routes in Multicast VPN
How a multicast ingress router learns which egress routers want each flow, with a single wildcard request.
Dolganow, A., Kotalwar, J., Rosen, E., Ed., and Z. Zhang, "Explicit Tracking with Wildcard Routes in Multicast VPN", RFC 8534, DOI 10.17487/RFC8534, February 2019, <https://www.rfc-editor.org/info/rfc8534>.
@misc{rfc8534,
series = {Request for Comments},
number = 8534,
howpublished = {RFC 8534},
publisher = {RFC Editor},
doi = {10.17487/RFC8534},
url = {https://www.rfc-editor.org/info/rfc8534},
author = {A. Dolganow and J. Kotalwar and E. Rosen and Z. Zhang},
title = {{Explicit Tracking with Wildcard Routes in Multicast VPN}},
pagetotal = 21,
year = 2019,
month = feb,
} RFC text
Contents
Internet Engineering Task Force (IETF) A. Dolganow
Request for Comments: 8534 J. Kotalwar
Updates: 6514, 6625, 7524, 7582, 7900 Nokia
Category: Standards Track E. Rosen, Ed.
ISSN: 2070-1721 Z. Zhang
Juniper Networks, Inc.
February 2019
Explicit Tracking with Wildcard Routes in Multicast VPN
Abstract
The base Multicast VPN (MVPN) specifications (RFCs 6513 and 6514)
provide procedures to allow a multicast ingress node to invoke
"explicit tracking" for a multicast flow or set of flows, thus
learning the egress nodes for that flow or set of flows. However,
the specifications are not completely clear about how the explicit
tracking procedures work in certain scenarios. This document
provides the necessary clarifications. It also specifies a new,
optimized explicit-tracking procedure. This new procedure allows an
ingress node, by sending a single message, to request explicit
tracking of each of a set of flows, where the set of flows is
specified using a wildcard mechanism. This document updates RFCs
6514, 6625, 7524, 7582, and 7900.
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
https://www.rfc-editor.org/info/rfc8534.
Copyright Notice
Copyright (c) 2019 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
(https://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. Terminology . . . . . . . . . . . . . . . . . . . . . . . 5
2. The Explicit-Tracking Flags . . . . . . . . . . . . . . . . . 5
3. Match for Tracking versus Match for Reception . . . . . . . . 7
4. Ingress Node Initiation of Tracking . . . . . . . . . . . . . 9
5. Egress Node Response to the Match for Tracking . . . . . . . 11
5.1. General Egress Node Procedures . . . . . . . . . . . . . 11
5.2. Responding to the LIR-pF Flag . . . . . . . . . . . . . . 12
5.3. When the Egress Node Is an ABR or ASBR . . . . . . . . . 15
6. Ingress Node Handling of Received Leaf A-D Routes with
LIR-pF Set . . . . . . . . . . . . . . . . . . . . . . . . . 16
7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17
8. Security Considerations . . . . . . . . . . . . . . . . . . . 18
9. References . . . . . . . . . . . . . . . . . . . . . . . . . 19
9.1. Normative References . . . . . . . . . . . . . . . . . . 19
9.2. Informative References . . . . . . . . . . . . . . . . . 20
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . 20
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 21
1. Introduction
The base Multicast VPN (MVPN) specifications, [RFC6513] and
[RFC6514], define the "Selective Provider Multicast Service Interface
Auto-Discovery route" (S-PMSI A-D route).
Per those RFCs, the S-PMSI A-D route contains a Network Layer
Reachability Information (NLRI) field that identifies a particular
multicast flow. In the terminology of those RFCs, each flow is
denoted by (C-S,C-G), where C-S is an IP source address and C-G is an
IP multicast address, both in the address space of a VPN customer.
The (C-S,C-G) of the multicast flow is encoded into the NLRI field.
An S-PMSI A-D route also carries a PMSI Tunnel attribute (PTA).
Typically, the PTA is used to identify a tunnel through the provider
backbone network (a "P-tunnel").
By originating an S-PMSI A-D route identifying a particular multicast
flow and a particular P-tunnel, a node is advertising the following:
If the node has to transmit packets of the identified flow over
the backbone, it will transmit them through the identified tunnel.
[RFC6513] and [RFC6514] also define a procedure that allows an
ingress node of a particular multicast flow to determine the set of
egress nodes that have requested to receive that flow from that
ingress node. The ability of an ingress node to identify the egress
nodes for a particular flow is known as "explicit tracking". An
ingress node requests explicit tracking by setting a flag (the "Leaf
Information Required" flag, or LIR flag) in the PTA. When an egress
node receives an S-PMSI A-D route with the LIR flag set, the egress
node originates a Leaf A-D route whose NLRI field contains the NLRI
from the corresponding S-PMSI A-D route. In this way, the egress
node advertises that it has requested to receive the particular flow
identified in the NLRI of that S-PMSI A-D route.
[RFC6513] and [RFC6514] also allow an ingress node to originate an
S-PMSI A-D route whose PTA has the LIR flag set but that does not
identify any P-tunnel. This mechanism can be used when desired to do
explicit tracking of a flow without at the same time binding that
flow to a particular P-tunnel.
[RFC6625] (and other RFCs that update it) extends the specification
of S-PMSI A-D routes and allows an S-PMSI A-D route to encode a
wildcard in its NLRI. Either the C-S or the C-G or both can be
replaced by wildcards. These routes are known as (C-*,C-S) S-PMSI
A-D routes, (C-S,C-*) S-PMSI A-D routes, or (C-*,C-*) S-PMSI A-D
routes, depending on whether the C-S or C-G or both have been
replaced by wildcards. These routes are known jointly as "wildcard
S-PMSI A-D routes".
One purpose of this document is to clarify the way that the explicit
tracking procedures of [RFC6513] and [RFC6514] are applied when
wildcard S-PMSI A-D routes are used.
In addition, this document addresses the following scenario, which is
not addressed in [RFC6513], [RFC6514], or [RFC6625]. Suppose an
ingress node originates an S-PMSI A-D route whose NLRI specifies, for
example, (C-*,C-*) (i.e., both C-S and C-G are replaced by wildcards)
and whose PTA identifies a particular P-tunnel. Now suppose that the
ingress node wants explicit tracking for each individual flow that it
transmits (following the procedures of [RFC6625]) on that P-tunnel.
In this example, if the ingress node sets the LIR flag in the PTA of
the wildcard S-PMSI A-D route, each egress node that needs to receive
a flow from the ingress node will respond with a Leaf A-D route whose
NLRI contains the (C-*,C-*) wildcard. This allows the ingress node
to determine the set of egress nodes that are interested in receiving
flows from the ingress node. However, it does not allow the ingress
node to determine exactly which flows are of interest to which egress
nodes.
If the ingress node needs to determine which egress nodes are
interested in receiving which flows, it needs to originate an S-PMSI
A-D route for each individual (C-S,C-G) flow that it is transmitting,
and it needs to set the LIR flag in the PTA of each such route.
However, since all the flows are being sent through the tunnel
identified in the (C-*,C-*) S-PMSI A-D route, there is no need to
identify a tunnel in the PTA of each (C-S,C-G) S-PMSI A-D route. Per
Section 5 of [RFC6514], the PTA of the (C-S,C-G) S-PMSI A-D routes
can specify "no tunnel information present". This procedure allows
explicit tracking of individual flows, even though all those flows
are assigned to tunnels by wildcard S-PMSI A-D routes.
However, this procedure requires several clarifications:
o The procedures of [RFC6625] do not address the case of an S-PMSI
A-D route whose NLRI contains wildcards but whose PTA specifies
"no tunnel information present".
o If it is desired to send a set of flows through the same tunnel
(where that tunnel is advertised in a wildcard S-PMSI A-D route),
but it is also desired to explicitly track each individual flow
transmitted over that tunnel, one has to send an S-PMSI A-D route
(with the LIR flag set in the PTA) for each individual flow. It
would be more optimal if the ingress node could just send a single
wildcard S-PMSI A-D route binding the set of flows to a particular
tunnel and have the egress nodes respond with Leaf A-D routes for
each individual flow.
o [RFC6513] and [RFC6514] support the notion of "segmented
P-tunnels", where "segmentation" occurs at Autonomous System
Border Routers (ASBRs); [RFC7524] extends the notion of segmented
P-tunnels so that segmentation can occur at Area Border Routers
(ABRs). One can think of a segmented P-tunnel as passing through
a number of "segmentation domains". In each segmentation domain,
a given P-tunnel has an ingress node and a set of egress nodes.
The explicit tracking procedures allow an ingress node of a
particular segmentation domain to determine, for a particular flow
or set of flows, the egress nodes of that segmentation domain.
This has given rise to two further problems:
* The explicit tracking procedures do not allow an ingress node
to "see" past the boundaries of the segmentation domain.
* The prior specifications do not make it very clear whether a
segmented tunnel egress node, upon receiving an S-PMSI A-D
route whose PTA specifies "no tunnel information present", is
expected to forward the S-PMSI A-D route, with the same PTA, to
the next segmentation domain.
These problems are addressed in Section 5.3.
This document clarifies the procedures for originating and receiving
S-PMSI A-D routes and Leaf A-D routes. This document also adds new
procedures to allow more efficient explicit tracking. The procedures
being clarified and/or extended are discussed in multiple places in
the documents being updated.
1.1. Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
"OPTIONAL" in this document are to be interpreted as described in
BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
capitals, as shown here.
2. The Explicit-Tracking Flags
[RFC6514] defines one flag in the PTA, the "Leaf Information
Required" (LIR) flag, that is used for explicit tracking.
This document defines a new flag in the Flags field of the PMSI
Tunnel attribute. This new flag is known as the "Leaf Information
Required per Flow" flag (LIR-pF). This flag may be set in the PTA of
a (C-*,C-*), (C-*,C-G), or (C-S,C-*) S-PMSI A-D route. The
conditions under which it should be set by the originator of the
route are discussed in Section 4. The significance of the flag in a
received wildcard S-PMSI A-D route is discussed in Sections 5 and
5.2.
The LIR-pF flag may also be set in the PTA of a Leaf A-D route. The
conditions under which it should be set by the originator of the
route are discussed in Section 5.2. The significance of the flag in
a received Leaf A-D route is discussed in Section 6.
Note that support for the LIR-pF flag is OPTIONAL. This flag SHOULD
NOT be set in a route's PTA unless it is known that the flag is
supported by all the Provider Edge (PE) routers that are to receive
that route. Typically, this might mean that the ability to set this
flag would be controlled by a configuration knob, and operators would
not set this knob unless they know that all the relevant PEs support
this feature. How this is known is outside the scope of this
document.
This document only defines procedures for the LIR-pF flag when that
flag is in the PTA of a wildcard S-PMSI A-D route or a Leaf A-D
route. In all other cases, the flag SHOULD be clear, and its value
SHOULD be ignored. Use of the flag in these other cases is outside
the scope of this document.
Section 5 of [RFC6514] lists a number of tunnel types. We will refer
to these as "6514-tunnel-types". Other tunnel types will be referred
to as "non-6514-tunnel-types". This document specifies procedures
for using the LIR-pF flag with 6514-tunnel-types. Procedures for
using the LIR-pF flag with non-6514-tunnel-types are outside the
scope of this document.
If it is desired to use a particular non-6514-tunnel-type in MVPN,
there needs to be a specification for how that tunnel type is used in
MVPN. If it is desired to use that tunnel type along with the LIR-pF
flag, that specification (or a follow-on specification) will have to
specify the rules for using the LIR-pF flag with that tunnel type.
As an example, see [BIER-MVPN]. In the absence of such a
specification (or in the absence of support for such a
specification):
o the originator of a route that carries a PTA SHOULD NOT set the
LIR-pF flag in any PTA that specifies that tunnel type, and
o the receiver of a route that carries a PTA specifying that tunnel
type SHOULD treat the LIR-pF flag as if it were not set.
If the LIR-pF flag is set in the PTA of an S-PMSI A-D route, the
originator of that route MUST also set the LIR flag. If the PTA of a
received wildcard S-PMSI A-D route has the LIR-pF flag set but does
not have the LIR flag set, the receiver MUST log the fact that the
flags appear to have been improperly set. However, the route MUST
then be processed normally (as if both flags were set), as specified
in this document.
It is worth noting what will happen if the LIR-pF flag is set in the
PTA of, for example, a (C-*,C-*) S-PMSI A-D route originated by an
ingress node, but one or more of the egress nodes do not support the
LIR-pF flag:
1. The ingress node will not be able to determine the complete set
of egress nodes that are expecting a particular multicast flow
from that ingress node.
2. Depending upon the tunnel type, the ingress node may send a
particular multicast flow only to the egress nodes that do
support the LIR-pF flag. From the perspective of egress nodes
that do not support the LIR-pF flag, certain flows may appear to
be "blackholed".
It is also worth noting that it is possible for an ingress node that
sets the LIR-pF flag in an S-PMSI A-D route to detect the presence of
egress nodes that do not support this flag.
Since an ingress node that sets the LIR-pF flag is also required to
set the LIR flag, egress nodes that do not support the LIR-pF flag
will respond, as specified in [RFC6514], to the ingress node's
(C-*,C-*) S-PMSI A-D route with a Leaf A-D route.
As discussed in Section 5.2, any Leaf A-D route originated in
response to an S-PMSI A-D route that has the LIR-pF flag set will
carry a PTA whose LIR-pF flag is set. If an ingress node receives a
Leaf A-D route whose Route Key field corresponds to the NLRI of an
S-PMSI A-D route whose PTA has the LIR-pF flag set, but the Leaf A-D
route lacks a PTA or has a PTA where the LIR-pF flag is clear, the
ingress node can infer that the egress node originating that Leaf A-D
route does not support the LIR-pF flag. The software at the ingress
node MUST detect this and MUST have a way of alerting the operator
that the deployment is not properly configured.
3. Match for Tracking versus Match for Reception
Section 3.2 of [RFC6625] specifies a set of rules for finding the
S-PMSI A-D route that is the "match for data reception" (or more
simply, the "match for reception") of a given (C-S,C-G) or (C-*,C-G)
state. These rules do not take into account the fact that some
S-PMSI A-D routes may not be carrying PTAs at all or may be carrying
PTAs that do not identify any P-tunnel. (A PTA that does not
identify any P-tunnel is one whose Tunnel Type field has been set to
"no tunnel information present", as specified in Section 5 of
[RFC6514].)
The rules for finding a "match for reception" in [RFC6625] are hereby
modified as follows:
When applying the rules of Sections 3.2.1 or 3.2.2 of [RFC6625],
it is REQUIRED to ignore any S-PMSI A-D route that has no PTA, or
whose PTA specifies "no tunnel information present".
There are other RFCs that update [RFC6625] and modify the rules for
finding a "match for reception". See, e.g., [RFC7582] and [RFC7900].
When applying those modified rules, it is REQUIRED to ignore any
S-PMSI A-D route that has no PTA, or whose PTA specifies "no tunnel
information present".
We also introduce a new notion, the "match for tracking":
For a given C-flow ((C-S,C-G) or (C-*,C-G)), the "match for
tracking" is chosen as follows. Ignore any S-PMSI A-D route that
has no PTA. Also ignore any S-PMSI A-D route whose PTA specifies
"no tunnel information present" and has neither the LIR flag nor
the LIR-pF flag set. (That is, *do not* ignore an S-PMSI A-D
route that has a PTA specifying "no tunnel information present"
unless its LIR and LIR-pF flags are both clear). Then apply the
rules (from [RFC6625] and other documents that update it) for
finding the "match for reception". The result (if any) is the
"match for tracking".
Note that the procedure for finding the match for tracking takes
into account S-PMSI A-D routes whose PTAs specify "no tunnel
information present" and that have either the LIR or LIR-pf flag
set. The procedure for finding the match for reception ignores
such routes.
We will clarify this with a few examples. In these examples, we
assume that there is only one segmentation domain. In this case, the
ingress and egress nodes are PE routers.
Suppose a given PE router, PE1, has chosen PE2 as the "upstream PE"
[RFC6513] for a given flow (C-S1,C-G1). And suppose PE1 has
installed the following two routes that were originated by PE2:
o Route1: A (C-*,C-*) S-PMSI A-D route whose PTA specifies a tunnel.
o Route2: A (C-S1,C-G1) S-PMSI A-D route whose PTA specifies "no
tunnel information present" and has the LIR flag set.
Route1 is the match of (C-S1,C-G1) for reception, and Route2 is the
match of (C-S1,C-G1) for tracking.
Continuing this example, suppose:
o PE1 has chosen PE2 as the upstream PE for a different flow,
(C-S2,C-G2).
o PE2 has not originated an S-PMSI A-D route for (C-S2,C-G2).
In this case, PE1 would consider Route1 to be the match of
(C-S2,C-G2) for tracking as well as its match for reception.
Also note that if a match for tracking does not have the LIR flag or
the LIR-pF flag set, no explicit tracking information will be
generated. See Section 5.
As another example, suppose PE1 has installed the following two
routes that were originated by PE2:
o Route1: A (C-*,C-*) S-PMSI A-D route (irrespective of whether the
PTA specifies a tunnel).
o Route2: A (C-S1,C-G1) S-PMSI A-D route whose PTA specifies a
tunnel.
In this case, Route2 is both the "match for reception" and the "match
for tracking" for (C-S1,C-G1).
Note that for a particular C-flow, PE1's match for reception might be
the same route as its match for tracking, or its match for reception
might be a "less specific" route than its match for tracking. But
its match for reception can never be a "more specific" route than its
match for tracking.
4. Ingress Node Initiation of Tracking
An ingress node that needs to initiate explicit tracking for a
particular flow or set of flows can do so by performing one of the
following procedures:
1. An ingress node can initiate explicit tracking for (C-S1,C-G1) by
originating an S-PMSI A-D route that identifies (C-S1,C-G1) in
its NLRI, including a PTA in that route, and setting the LIR flag
in that PTA. The PTA may specify either a particular tunnel or
"no tunnel information present".
However, the PTA of the (C-S1,C-G1) S-PMSI A-D route SHOULD NOT
specify "no tunnel information present" unless the ingress node
also originates an A-D route carrying a PTA that specifies the
tunnel to be used for carrying (C-S1,C-G1) traffic. Such a route
could be an "Inclusive Provider Multicast Service Interface Auto-
Discovery route" (I-PMSI A-D route), a (C-*,C-G1) S-PMSI A-D
route, a (C-S1,C-*) S-PMSI A-D route, or a (C-*,C-*) S-PMSI A-D
route. (There is no point in requesting explicit tracking for a
given flow if there is no tunnel on which the flow is being
carried.)
Note that if the ingress node originates a wildcard S-PMSI A-D
route carrying a PTA specifying the tunnel to be used for
carrying (C-S1,C-G1) traffic, and if that PTA has the LIR-pF flag
set, then explicit tracking for (C-S1,C-G1) is requested by that
S-PMSI A-D route. In that case, the ingress node SHOULD NOT
originate a (C-S1,C-G1) S-PMSI A-D route whose PTA specifies "no
tunnel information present"; such a route would not provide any
additional functionality.
To terminate explicit tracking that has been initiated by an
S-PMSI A-D route whose PTA specifies "no tunnel information
present", the ingress node withdraws the route.
To terminate explicit tracking that has been initiated by an
S-PMSI A-D route whose PTA specifies a tunnel, the ingress node
re-originates the route without the LIR flag set.
2. The following procedure can be used if and only if it is known
that the egress nodes support the optional LIR-pF flag. If the
ingress node originates a wildcard S-PMSI A-D route, it can
initiate explicit tracking for the individual flows that match
the wildcard route by setting the LIR-pF flag in the PTA of the
wildcard route. If an egress node needs to receive one or more
flows for which that wildcard route is a match for tracking, the
egress node will originate a Leaf A-D route for each such flow,
as specified in Section 5.2).
When following this procedure, the PTA of the S-PMSI A-D route
may specify either a tunnel or "no tunnel information present".
The choice between these two options is determined by
considerations that are outside the scope of this document.
To terminate explicit tracking that has been initiated by an
S-PMSI A-D route whose PTA specifies "no tunnel information
present", the ingress node withdraws the route.
To terminate explicit tracking that has been initiated by an
S-PMSI A-D route whose PTA specifies a tunnel, the ingress node
re-originates the route without either the LIR or LIR-pF flags
set.
Note that this procedure (Procedure 2 of Section 4) may not yield
the expected results if there are egress nodes that do not
support the LIR-pF flag; hence, it SHOULD NOT be used in that
case.
5. Egress Node Response to the Match for Tracking
5.1. General Egress Node Procedures
There are four cases to consider:
1. With regard to a particular (C-S,C-G) or (C-*,C-G) multicast
state, the egress node's match for tracking is the same as its
match for reception, and neither the LIR nor the LIR-pF flags are
set.
In this case, the egress node does not originate a Leaf A-D route
in response to the match for reception/tracking, and there is no
explicit tracking of the flow. This document specifies no new
procedures for this case.
2. With regard to a particular (C-S,C-G) or (C-*,C-G) multicast
state, the egress node's match for tracking is the same as its
match for reception, and the LIR flag is set, but the LIR-pF flag
is not set.
In this case, a Leaf A-D route is originated by the egress node,
corresponding to the S-PMSI A-D route that is the match for
reception/tracking. Construction of the Leaf A-D route is as
specified in [RFC6514]; this document specifies no new procedures
for this case.
3. With regard to a particular (C-S,C-G) or (C-*,C-G) multicast
state, the egress node's match for tracking is the same as its
match for reception, and LIR-pF is set. The egress node follows
whatever procedures are required by other specifications, based
on the match for reception. However, any Leaf A-D route
originated by the egress node as a result MUST have the LIR-pF
flag set in its PTA. The egress node MUST also follow the
procedures of Section 5.2.
4. With regard to a particular (C-S,C-G) or (C-*,C-G) multicast
state, the egress node's match for tracking is *not* the same as
its match for reception. This can only happen if the match for
tracking has a PTA specifying "no tunnel information present",
with either the LIR flag or the LIR-pF flag set. In this case,
the egress node MUST respond, separately, to *both* the match for
tracking and the match for reception.
If a Leaf A-D route is originated in response to the match for
reception, the LIR-pF flag in the Leaf A-D route's PTA MUST have
the same value as the LIR-pF flag in the match for reception's
PTA. In all other respects, the procedures for responding to the
match for reception are not affected by this document.
If the match for tracking has the LIR flag set but the LIR-pF
flag is not set, then the behavior of the egress node is not
affected by the procedures of this document.
If the match for tracking has the LIR-pF flag set, the egress
node MUST follow the procedures of Section 5.2.
Note that if the LIR flag is set in the PTA of the match for
reception, the egress node may need to originate one or more Leaf
A-D routes corresponding to the match for tracking, as well as
originating a Leaf A-D route corresponding to the match for
reception.
5.2. Responding to the LIR-pF Flag
To respond to a match for tracking that has the LIR-pF flag set, an
egress node originates one or more Leaf A-D routes.
Suppose the egress node has multicast state for a (C-S,C-G) or a
(C-*,C-G) flow and has determined a particular S-PMSI A-D route,
which has the LIR-pF flag set, to be the match for tracking for that
flow. Then if the egress node supports the LIR-pF flag, it MUST
originate a Leaf A-D route whose NLRI identifies that particular
flow. Note that if a single S-PMSI A-D route (with wildcards) is the
match for tracking for multiple flows, the egress node may need to
originate multiple Leaf A-D routes, one for each such flow. We say
that, from the perspective of a given egress node, a given S-PMSI A-D
route tracks the set of flows for which it is the match for tracking.
Each of the Leaf A-D routes originated in response to that S-PMSI A-D
route tracks a single such flow.
The NLRI of each Leaf A-D route that tracks a particular flow is
constructed as follows. The Route Key field of the NLRI will have
the format shown in Figure 1 (as defined in Sections 4.3 and 4.4 of
[RFC6514]).
+------------------------------------+
| RD (8 octets) |
+------------------------------------+
| Multicast Source Length (1 octet) |
+------------------------------------+
| Multicast Source (Variable) |
+------------------------------------+
| Multicast Group Length (1 octet) |
+------------------------------------+
| Multicast Group (Variable) |
+------------------------------------+
| Ingress PE's IP Address |
+------------------------------------+
Figure 1: NLRI of S-PMSI A-D Route
o The "ingress PE" address is taken from the Originating Router
field of the NLRI of the S-PMSI A-D route that is the match for
tracking. Section 2 of [RFC6515] explains how the receiver of a
Leaf A-D route determines the length of this field and the address
family of the PE's IP address.
o The Multicast Source and Multicast Group fields respectively
specify a source address (S) and a group address(G) that together
identify the flow or flows being tracked by this Leaf A-D route.
If a (C-*,C-G) is being tracked by this Leaf A-D route, the
Multicast Source field is omitted, and the Multicast Source Length
field is set to 0. In this case, the Leaf A-D route is known as a
"wildcard Leaf A-D route".
o The Route Distinguisher (RD) field is set to the value of the RD
field from the NLRI of the S-PMSI A-D route.
The encoding of these Leaf A-D routes is similar to the encoding of
the Leaf A-D routes described in Section 6.2.2 of [RFC7524], which
were designed for the support of "global table multicast". However,
that document sets the RD to either 0 or -1; following the procedures
of the present document, the RD will never be 0 or -1. Therefore,
Leaf A-D routes constructed according to the procedures of this
section can always be distinguished from the Leaf A-D routes
constructed according to the procedures of Section 6.2.2 of
[RFC7524]. Also, Leaf A-D routes constructed according to the
procedures of this section are VPN-specific routes and will always
carry an IP-address-specific Route Target, as specified in [RFC6514].
If a Leaf A-D route is originated as a response to a match for
tracking whose PTA specifies "no tunnel information present", the
Leaf A-D route MUST carry a PTA that specifies "no tunnel information
present". The LIR-pF flag in this PTA MUST be set.
If an egress node originates multiple Leaf A-D routes in response to
a single S-PMSI A-D route, and that S-PMSI A-D route is later
withdrawn, then those Leaf A-D routes MUST also be withdrawn.
Similarly, a Leaf A-D route needs to be withdrawn (either implicitly
or explicitly) if the egress node changes its Upstream Multicast Hop
(UMH) [RFC6513] for the flow that is identified in the Leaf A-D
route's NLRI, or if the egress node that originated the route no
longer needs to receive that flow.
It is possible that an egress node will acquire (C-S,C-G) state or
(C-*,C-G) state after it has already received the S-PMSI A-D that is
the match for tracking for that state. In this case, a Leaf A-D
route needs to be originated at that time, and the egress node must
remember that the new Leaf A-D route corresponds to that match for
tracking.
If a particular S-PMSI A-D route is a match for tracking but not a
match for reception, the LIR flag in its PTA is ignored if the LIR-pF
flag is set.
When the match for tracking is the same as the match for reception,
the PTA of the match for tracking/reception will have specified a
tunnel type. Some of the rules for constructing the PTA of the Leaf
A-D route depend on the tunnel type, and some are independent of the
tunnel type. No matter what the tunnel type is, the LIR-pF flag MUST
be set.
If the match for tracking/reception is a wildcard S-PMSI A-D route,
the egress node may originate a wildcard Leaf A-D route in response,
as well as originating one or more non-wildcard Leaf A-D routes.
Note that the LIR-pF flag MUST be set in the wildcard Leaf A-D route
as well as in the non-wildcard Leaf A-D routes.
This document provides additional rules for constructing the PTA when
the tunnel type is a 6514-tunnel-type (see Section 2).
As discussed in Section 2, if a non-6514-tunnel-type is being used,
then presumably there is a specification for how that tunnel type is
used in MVPN. If it is desired to use that tunnel type along with
the LIR-pF flag, that specification (or a follow-on specification)
will have to specify the additional rules for constructing the PTA.
As an example, see [BIER-MVPN].
For 6514-tunnel-types, additional rules for constructing the PTA are
as follows:
o If the tunnel type of the PTA attached to the match for tracking/
reception is Ingress Replication, the Leaf A-D route's PTA MAY
specify Ingress Replication. In this case, the MPLS Label field
of the PTA MAY be a non-zero value. If so, this label value will
be used by the ingress PE when it transmits, to the egress PE,
packets of the flow identified in the Leaf A-D route's NLRI.
Alternatively, the egress PE MAY specify an MPLS label value of
zero, or it MAY specify a tunnel type of "no tunnel information
present". In either of these cases, when the ingress PE transmits
packets of the identified flow to the egress PE, it will use the
label that the egress PE specified in the PTA of the Leaf A-D
route that it originated in response to the LIR flag of the match
for reception.
o If the tunnel type of the PTA attached to the match for tracking/
reception is any of the other 6514-tunnel-types, the PTA attached
to the Leaf A-D route MUST specify a tunnel type of "no tunnel
information present".
It may happen that the tunnel type is a non-6514-tunnel type, but
either (a) there is no specification for how to use that tunnel type
with the LIR-pF flag or (b) there is such a specification, but the
egress node does not support it. In that case, the egress node MUST
treat the match for tracking/reception as if it had the LIR-pF flag
clear.
5.3. When the Egress Node Is an ABR or ASBR
When segmented P-tunnels are used, the ingress and egress nodes may
be ABRs or ASBRs. An egress ABR/ASBR that receives and installs an
S-PMSI A-D route also forwards that route. If the received PTA of an
installed S-PMSI A-D route specifies a tunnel, the egress ABR/ASBR
MAY change the PTA before forwarding the route, in order to specify a
different tunnel type (as discussed in [RFC6514] and/or [RFC7524]).
The egress ABR/ASBR may also need to originate a Leaf A-D route, as
specified in [RFC6514] and/or [RFC7524].
Suppose the S-PMSI A-D route as received has a PTA specifying a
tunnel and also has the LIR-pF flag set. The egress ABR/ASBR
originates a corresponding Leaf A-D route for a given (C-S,C-G) only
if it knows that it needs to receive that flow. It will know this by
virtue of receiving a corresponding Leaf A-D route from downstream.
(In the case where the PTA specifies a tunnel but the LIR-pF flag is
not set, this document does not specify any new procedures.)
The procedures in the remainder of this section apply only when an
egress ABR/ASBR has installed an S-PMSI A-D route whose PTA as
received specifies "no tunnel information present" but has the LIR
flag or the LIR-pF flag set.
If the received PTA of the installed S-PMSI A-D route specifies "no
tunnel information present", the egress ABR/ASBR MUST pass the PTA
along unchanged when it forwards the S-PMSI A-D route. (That is, a
PTA specifying "no tunnel information present" MUST NOT be changed
into a PTA specifying a tunnel.) Furthermore, if the PTA specifies
"no tunnel information present", the LIR and LIR-pF flags in the PTA
MUST be passed along unchanged.
As a result of propagating such an S-PMSI A-D route, the egress ABR/
ASBR may receive one or more Leaf A-D routes that correspond to that
S-PMSI A-D route. These routes will be received carrying an
IP-address-specific Route Target (RT) Extended Community that
specifies the address of the egress ABR/ASBR. The egress ABR/ASBR
will propagate these Leaf A-D routes after changing the RT as
follows. The Global Administrator field of the modified RT will be
set to the IP address taken either from the S-PMSI A-D route's
Next-Hop field [RFC6514] or its Segmented Point-to-Multipoint (P2MP)
Next-Hop Extended Community [RFC7524]. The address from the
Segmented P2MP Next-Hop Extended Community is used if that Extended
Community is present; otherwise, the address from the Next-Hop field
is used.
This procedure enables the ingress PE to explicitly track the egress
PEs for a given flow, even if segmented tunnels are being used.
However, cross-domain explicit tracking utilizes S-PMSI A-D routes
that do not specify tunnel information; therefore, it can only be
done when the S-PMSI A-D route that is a flow's match for tracking is
different from the S-PMSI A-D route that is that flow's match for
reception.
6. Ingress Node Handling of Received Leaf A-D Routes with LIR-pF Set
Consider the following situation:
o An ingress node, call it N, receives a Leaf A-D route, call it L.
o L carries an IP-address-specific RT identifying N.
o The Route Key field of L's NLRI is not identical to the NLRI of
any current I-PMSI or S-PMSI A-D route originated by N.
Per the procedures of [RFC6514] and [RFC7524], such a Leaf A-D route
does not cause any MVPN-specific action to be taken by N.
This document modifies those procedures in the case where there is a
current wildcard S-PMSI A-D route, originated by N, to which L is a
valid response according to the procedures of Section 5.2. In this
case, L MUST be processed by N.
Suppose that L's PTA specifies a tunnel type of Ingress Replication
and that it also specifies a non-zero MPLS label. Then if N needs to
send to L a packet belonging to the multicast flow or flows
identified in L's NLRI, N MUST use the specified label.
If L's PTA meets any of the following conditions:
o It specifies a tunnel type of "no tunnel information present", or
o It specifies a tunnel type of Ingress Replication, but specifies
an MPLS label of zero, or
o It specifies any other 6514-tunnel-type,
then the action taken by N when it receives L is a local matter. In
this case, the Leaf A-D route L provides N with explicit tracking
information for the flow identified by L's NLRI. However, that
information is for management/monitoring purposes and does not have
any direct effect on the flow of multicast traffic.
If L's PTA specifies a non-6514-tunnel-type not mentioned above,
presumably there is a specification for how MVPN uses that tunnel
type. If the LIR-pF flag is to be used with that tunnel type, that
specification must specify the actions that N is to take upon
receiving L. As an example, see [BIER-MVPN]. In the absence of such
a specification, the LIR-pF flag SHOULD BE ignored. See Section 2
for further discussion of non-6514-tunnel-types.
7. IANA Considerations
IANA has added the following entry to the "P-Multicast Service
Interface (PMSI) Tunnel Attribute Flags" registry under the "Border
Gateway Protocol (BGP) Parameters" registry. This registry is
defined in [RFC7902]. The entry appears as:
o Value: 2
o Name: Leaf Information Required per-Flow (LIR-pF)
o Reference: RFC 8534
8. Security Considerations
The Security Considerations of [RFC6513] and [RFC6514] apply.
By setting the LIR-pF flag in a single wildcard S-PMSI A-D route, a
large number of Leaf A-D routes can be elicited. If this flag is set
when not desired (through either error or malfeasance), a significant
increase in control plane overhead can result. Properly protecting
the control plane should prevent this kind of attack.
In the event such an attack occurs, mitigating it is unfortunately
not very straightforward. The ingress node can take note of the fact
that it is getting, in response to an S-PMSI A-D route that has
LIR-pF clear, one or more Leaf A-D routes that have LIR-pF set. By
default, the reception of such a route MUST be logged. However, it
is possible for such log entries to be "false positives" that
generate a lot of "noise" in the log; therefore, implementations
SHOULD have a knob to disable this logging.
In theory, if one or more Leaf A-D routes with LIR-pF set arrive in
response to an S-PMSI A-D route with LIR-pF clear, withdrawing the
S-PMSI A-D route could put a stop to the attack. In practice, that
is not likely to be a very good strategy, because:
o Under normal operating conditions, there are some race conditions
that may cause the ingress node to think it is being attacked,
when in fact it is not.
o If some egress nodes have a bug that causes them to set LIR-pF
when it should be clear, withdrawing the S-PMSI A-D route will
stop the flow of multicast data traffic to all the egress nodes,
causing an unnecessary customer-visible disruption.
o The same situation that caused the S-PMSI A-D route to be
originated in the first place will still exist after the S-PMSI
A-D route is withdrawn, so the route will just be re-originated.
In other words, any action that would ameliorate the effects of this
sort of attack would likely have a negative effect during normal
operation. Therefore, it is really better to rely on security
mechanisms that protect the control plane generally than to have a
mechanism that is focused on this one particular type of attack.
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,
<https://www.rfc-editor.org/info/rfc2119>.
[RFC6513] Rosen, E., Ed. and R. Aggarwal, Ed., "Multicast in MPLS/
BGP IP VPNs", RFC 6513, DOI 10.17487/RFC6513, February
2012, <https://www.rfc-editor.org/info/rfc6513>.
[RFC6514] Aggarwal, R., Rosen, E., Morin, T., and Y. Rekhter, "BGP
Encodings and Procedures for Multicast in MPLS/BGP IP
VPNs", RFC 6514, DOI 10.17487/RFC6514, February 2012,
<https://www.rfc-editor.org/info/rfc6514>.
[RFC6515] Aggarwal, R. and E. Rosen, "IPv4 and IPv6 Infrastructure
Addresses in BGP Updates for Multicast VPN", RFC 6515,
DOI 10.17487/RFC6515, February 2012,
<https://www.rfc-editor.org/info/rfc6515>.
[RFC6625] Rosen, E., Ed., Rekhter, Y., Ed., Hendrickx, W., and
R. Qiu, "Wildcards in Multicast VPN Auto-Discovery
Routes", RFC 6625, DOI 10.17487/RFC6625, May 2012,
<https://www.rfc-editor.org/info/rfc6625>.
[RFC7524] Rekhter, Y., Rosen, E., Aggarwal, R., Morin, T.,
Grosclaude, I., Leymann, N., and S. Saad, "Inter-Area
Point-to-Multipoint (P2MP) Segmented Label Switched Paths
(LSPs)", RFC 7524, DOI 10.17487/RFC7524, May 2015,
<https://www.rfc-editor.org/info/rfc7524>.
[RFC7902] Rosen, E. and T. Morin, "Registry and Extensions for
P-Multicast Service Interface Tunnel Attribute Flags",
RFC 7902, DOI 10.17487/RFC7902, June 2016,
<https://www.rfc-editor.org/info/rfc7902>.
[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
May 2017, <https://www.rfc-editor.org/info/rfc8174>.
9.2. Informative References
[BIER-MVPN]
Rosen, E., Sivakumar, M., Aldrin, S., Dolganow, A., and
T. Przygienda, "Multicast VPN Using BIER", Work in
Progress, draft-ietf-bier-mvpn-11, March 2018.
[RFC7582] Rosen, E., Wijnands, IJ., Cai, Y., and A. Boers,
"Multicast Virtual Private Network (MVPN): Using
Bidirectional P-Tunnels", RFC 7582, DOI 10.17487/RFC7582,
July 2015, <https://www.rfc-editor.org/info/rfc7582>.
[RFC7900] Rekhter, Y., Ed., Rosen, E., Ed., Aggarwal, R., Cai, Y.,
and T. Morin, "Extranet Multicast in BGP/IP MPLS VPNs",
RFC 7900, DOI 10.17487/RFC7900, June 2016,
<https://www.rfc-editor.org/info/rfc7900>.
Acknowledgments
The authors wish to thank Robert Kebler for his ideas and comments
and Stephane Litkowski and Benjamin Kaduk for their thorough reviews
and useful suggestions. We would also like to thank Mirja Kuhlewind
for her attention to the Security Considerations section.
Authors' Addresses
Andrew Dolganow
Nokia
438B Alexandra Rd #08-07/10
Alexandra Technopark
119968
Singapore
Email: andrew.dolganow@nokia.com
Jayant Kotalwar
Nokia
701 East Middlefield Rd
Mountain View, California 94043
United States of America
Email: jayant.kotalwar@nokia.com
Eric C. Rosen (editor)
Juniper Networks, Inc.
10 Technology Park Drive
Westford, Massachusetts 01886
United States of America
Email: erosen52@gmail.com
Zhaohui Zhang
Juniper Networks, Inc.
10 Technology Park Drive
Westford, Massachusetts 01886
United States of America
Email: zzhang@juniper.net
ISO Plain Language
On this page
§1The Problem With Wildcard Routes
The base Multicast VPN specifications, RFC 6513 and RFC 6514, define the Selective Provider Multicast Service Interface Auto-Discovery route, the S-PMSI A-D route. Its Network Layer Reachability Information field identifies a particular multicast flow, denoted by a pair of addresses: C-S, the IP source address, and C-G, the IP multicast address, both in the address space of a VPN customer. The route also carries a PMSI Tunnel attribute, typically used to identify a tunnel through the provider backbone, a P-tunnel. Originating such a route says that if the node has to transmit packets of the identified flow over the backbone, it will send them through the identified tunnel.
The same specifications define a procedure that lets the ingress node of a flow determine the set of egress nodes that have asked to receive that flow from it. That ability is called explicit tracking. The ingress node requests it by setting the Leaf Information Required flag, the LIR flag, in the PMSI Tunnel attribute. An egress node that receives an S-PMSI A-D route with the LIR flag set originates a Leaf A-D route whose NLRI contains the NLRI of the S-PMSI A-D route, advertising that it wants to receive that particular flow.
RFC 6625 extends the S-PMSI A-D route so that the source address, the group address or both can be replaced by wildcards, giving wildcard S-PMSI A-D routes. One purpose of RFC 8534 is to clarify how explicit tracking applies when wildcard routes are used.
The document also addresses a case the earlier specifications do not. Suppose the ingress node originates a (C-*,C-*) wildcard route that identifies a P-tunnel, and wants explicit tracking for each individual flow it transmits on that tunnel. If it sets only the LIR flag, each egress node answers with a Leaf A-D route carrying the (C-*,C-*) wildcard, which tells the ingress node which egress nodes want flows from it but not which flow each of them wants. To learn that, the ingress node has to originate an S-PMSI A-D route for every individual flow, with the LIR flag set in each, and since the tunnel is already fixed by the wildcard route, each of those routes can carry 'no tunnel information present'.
That works but needs clarifications, and it is not the most efficient form. It would be better if the ingress node could send a single wildcard route binding the set of flows to a tunnel and have the egress nodes answer with a Leaf A-D route for each individual flow. The RFC also addresses segmentation, where a P-tunnel passes through a number of segmentation domains, each with its own ingress node and set of egress nodes: the explicit tracking procedures do not let an ingress node see past the boundary of its domain, and the earlier specifications are unclear about whether a segmented tunnel egress node forwards an S-PMSI A-D route whose PTA says 'no tunnel information present' to the next domain.
Each router at the edge of a provider network advertises which customer multicast flows it wants. The router that receives the customer's traffic has to collect those answers to know where to send it. When a single advertisement covers many flows at once, the answers come back aggregated, and the sender learns who is interested but not in which flow. This RFC adds a way to get the finer answer without asking about every flow separately.
§2The Two Tracking Flags
RFC 6514 defines one flag in the PMSI Tunnel attribute, the Leaf Information Required flag, used for explicit tracking. This document defines a new flag in the Flags field of the same attribute, the Leaf Information Required per Flow flag, LIR-pF.
The LIR-pF flag may be set in the PTA of a (C-*,C-*), (C-*,C-G) or (C-S,C-*) wildcard S-PMSI A-D route, and it may also be set in the PTA of a Leaf A-D route. The flag is only defined for those cases; in all other cases it should be clear and its value should be ignored.
Support for the LIR-pF flag is optional. The flag should not be set in a route's PTA unless it is known that the flag is supported by all the Provider Edge routers that are to receive that route. Typically this means an operator would control the ability to set the flag with a configuration knob, and would not turn it on unless all the relevant routers support it.
If the LIR-pF flag is set in the PTA of an S-PMSI A-D route, the originator must also set the LIR flag. If the PTA of a received wildcard S-PMSI A-D route has LIR-pF set but not LIR, the receiver must log that the flags appear to have been set improperly, but must then process the route normally, as if both flags were set.
The RFC also limits itself to the tunnel types listed in Section 5 of RFC 6514, which it calls 6514-tunnel-types. Other tunnel types need their own specification to say how the LIR-pF flag is used with them. Until such a specification exists, the originator of a route should not set the flag in a PTA naming such a tunnel type, and the receiver should treat the flag as if it were not set.
If the flag is set but some egress nodes do not support it, two things follow. The ingress node cannot determine the complete set of egress nodes expecting a particular flow, and, depending on the tunnel type, it may send a flow only to the egress nodes that do support the flag, so that from the point of view of the others some flows appear to be blackholed. The ingress node can detect the problem: an egress node that does not support the flag answers with a Leaf A-D route whose PTA lacks the flag, and the software at the ingress node must detect this and must be able to alert the operator that the deployment is not properly configured.
IANA has registered the new flag with the value 2 in the PMSI Tunnel Attribute Flags registry.
There are now two flags in the tunnel attribute. The older one asks the receiver to answer at all. The new one asks for an answer per flow. The new flag is optional in a router, so an operator should only turn it on where every router understands it, and the ingress node can spot a router that does not and warn about it.
- It needs the older flagSetting LIR-pF on an S-PMSI A-D route requires setting LIR as well.
- Only two placesThe flag is defined for wildcard S-PMSI A-D routes and for Leaf A-D routes; elsewhere it should be clear.
- Support is optionalAn operator should only set it where every receiving router understands it.
- The sender can spot the gapAn answer without the flag tells the ingress node that a router did not support it, and it must be able to warn the operator.
§3Match for Tracking versus Match for Reception
Section 3.2 of RFC 6625 gives rules for finding the S-PMSI A-D route that is the match for data reception of a given state. Those rules did not take into account routes that carry no PTA at all, or a PTA that identifies no P-tunnel. This RFC modifies them: it is required to ignore any S-PMSI A-D route that has no PTA, or whose PTA specifies 'no tunnel information present'. The same requirement applies to the modified rules in other documents such as RFC 7582 and RFC 7900.
The RFC then introduces a new notion, the match for tracking. For a given flow, the match for tracking is chosen by ignoring any S-PMSI A-D route that has no PTA, and ignoring any S-PMSI A-D route whose PTA says 'no tunnel information present' and has neither the LIR flag nor the LIR-pF flag set. In other words such a route is ignored only when both flags are clear. The rules for finding the match for reception are then applied, and the result, if any, is the match for tracking.
The RFC works through examples. Suppose PE1 has chosen PE2 as the upstream PE for the flow (C-S1,C-G1) and has installed two routes originated by PE2: Route1, a (C-*,C-*) S-PMSI A-D route whose PTA specifies a tunnel, and Route2, a (C-S1,C-G1) S-PMSI A-D route whose PTA says 'no tunnel information present' and has the LIR flag set. Route1 is the match for reception and Route2 is the match for tracking. If PE1 has instead chosen PE2 as upstream for a different flow (C-S2,C-G2) and PE2 has not originated a route for it, then Route1 is the match for tracking for that flow as well as the match for reception.
If a match for tracking has neither the LIR flag nor the LIR-pF flag set, no explicit tracking information is generated. In another example, where Route1 is a (C-*,C-*) route and Route2 is a (C-S1,C-G1) route whose PTA specifies a tunnel, Route2 is both the match for reception and the match for tracking. The match for reception can be the same route as the match for tracking, or a less specific one, but it can never be more specific.
A router works out two different things about the routes it has been sent: which route says where the traffic should actually go, and which route is the one to answer. The two are usually the same. They differ when the route that names the tunnel is a broad one and a narrower route was sent just to ask for tracking. The RFC spells out how to pick each one.
| Route installed by PE1 | Match for reception | Match for tracking |
|---|---|---|
| (C-*,C-*) route whose PTA specifies a tunnel | Yes | Yes, when it is the only match |
| (C-S1,C-G1) route, no tunnel, LIR set | Ignored | Yes |
| (C-S1,C-G1) route whose PTA specifies a tunnel | Yes | Yes |
§4How an Ingress Node Starts Tracking
An ingress node that needs explicit tracking for a flow or a set of flows can use one of two procedures.
The first is to originate an S-PMSI A-D route that identifies the flow in its NLRI, including a PTA, and set the LIR flag in that PTA. The PTA may specify a particular tunnel or 'no tunnel information present'. However, the PTA of such a route should not say 'no tunnel information present' unless the ingress node also originates a route carrying a PTA that specifies the tunnel to be used for the traffic, which could be an Inclusive Provider Multicast Service Interface Auto-Discovery route, a (C-*,C-G1) route, a (C-S1,C-*) route or a (C-*,C-*) route. There is no point in requesting explicit tracking for a flow that is not being carried on any tunnel.
If the ingress node originates a wildcard S-PMSI A-D route carrying the PTA that specifies the tunnel, and that PTA has the LIR-pF flag set, then explicit tracking for the flow is requested by that wildcard route, and the ingress node should not originate a separate (C-S1,C-G1) route with 'no tunnel information present', since it would add nothing.
Tracking that was started by a route whose PTA says 'no tunnel information present' is terminated by withdrawing the route. Tracking that was started by a route whose PTA specifies a tunnel is terminated by re-originating the route with the LIR flag cleared.
The second procedure may be used only when it is known that the egress nodes support the optional LIR-pF flag. The ingress node originates a wildcard S-PMSI A-D route and sets the LIR-pF flag in its PTA, and the egress nodes that need any of the flows matching that wildcard each originate a Leaf A-D route for the flows they want. The PTA of the wildcard route may specify a tunnel or 'no tunnel information present'; the choice is outside the scope of the document. Tracking started this way is terminated by withdrawing the route, or, when the PTA specifies a tunnel, by re-originating the route with both the LIR and LIR-pF flags cleared.
This second procedure may not give the expected results if some egress nodes do not support the LIR-pF flag, so it should not be used in that case.
There are two ways to ask. The older way asks about one flow at a time. The newer way asks with a single broad route and asks the receivers to answer per flow, but it only works if every receiver understands the new flag. Both ways are undone in the same style, either by withdrawing the route or by sending it again with the flag turned off.
- S-PMSI A-D ROUTE, LIR AND LIR-pF SETOne wildcard route asks every egress router to answer for each flow it wants.
- LEAF A-D ROUTE PER FLOWOne reply for each flow this router wants, each naming the flow exactly.
§5.2How an Egress Node Answers
To respond to a match for tracking that has the LIR-pF flag set, an egress node originates one or more Leaf A-D routes. If the node has multicast state for a flow and has determined that a particular S-PMSI A-D route with the flag set is the match for tracking for that flow, then, if the node supports the flag, it must originate a Leaf A-D route whose NLRI identifies that particular flow. If a single wildcard S-PMSI A-D route is the match for tracking for several flows, the egress node may need to originate several Leaf A-D routes, one for each such flow.
The NLRI of each such Leaf A-D route is built from the fields shown in the RFC's Figure 1: the Route Distinguisher, the Multicast Source Length, the Multicast Source, the Multicast Group Length, the Multicast Group and the ingress PE's IP address. The ingress PE address is taken from the Originating Router field of the match for tracking. The source and group fields together identify the flow being tracked; if a (C-*,C-G) is being tracked the Multicast Source field is omitted and its length is set to 0, which makes the route a wildcard Leaf A-D route. The Route Distinguisher is copied from the NLRI of the match for tracking.
Because the RD in these routes is never 0 or -1, they can always be told apart from the Leaf A-D routes constructed for global table multicast, and they are VPN-specific routes that always carry an IP-address-specific Route Target.
If the Leaf A-D route is originated in response to a match for tracking whose PTA says 'no tunnel information present', the Leaf A-D route must carry a PTA that says 'no tunnel information present' with the LIR-pF flag set. If an egress node originates several Leaf A-D routes in response to a single S-PMSI A-D route, and that route is later withdrawn, the Leaf A-D routes must also be withdrawn. A Leaf A-D route also needs to be withdrawn if the egress node changes its upstream multicast hop for the flow, or no longer needs that flow.
For the tunnel types listed in RFC 6514 there are two extra rules. If the tunnel type of the match for tracking or reception is Ingress Replication, the Leaf A-D route's PTA may specify Ingress Replication, in which case the MPLS label field may be a non-zero value, and that label will be used by the ingress PE when it transmits to the egress PE the packets of the flow identified in the Leaf A-D route's NLRI. The egress PE may instead specify a label of zero, or specify 'no tunnel information present', in which case the ingress PE uses the label the egress PE gave in the PTA of the Leaf A-D route it originated in response to the LIR flag of the match for reception. For any other 6514 tunnel type, the PTA attached to the Leaf A-D route must specify 'no tunnel information present'.
The receiver answers with one route per flow it wants, each naming the flow exactly. The reply carries the flag so that the sender knows the answer is the detailed kind. If the reply names a label, the sender uses that label when it sends the traffic; otherwise it keeps using the label it already had.
RD (8 octets)
Multicast Source Length (1 octet)
Multicast Source (Variable)
Multicast Group Length (1 octet)
Multicast Group (Variable)
Ingress PE's IP Address§5.3When the Egress Node Is a Border Router
When segmented P-tunnels are used, the ingress and egress nodes may be Area Border Routers or Autonomous System Border Routers. An egress ABR or ASBR that receives and installs an S-PMSI A-D route also forwards that route. If the received PTA specifies a tunnel, the egress ABR or ASBR may change the PTA before forwarding the route in order to specify a different tunnel type.
If the received route has a PTA specifying a tunnel and the LIR-pF flag set, the egress ABR or ASBR originates a corresponding Leaf A-D route for a given flow only if it knows it needs to receive that flow, which it knows by having received a corresponding Leaf A-D route from downstream.
The rest of the section applies only when the egress ABR or ASBR has installed an S-PMSI A-D route whose PTA says 'no tunnel information present' but has the LIR flag or the LIR-pF flag set. In that case the PTA must be passed along unchanged when the route is forwarded: a PTA saying 'no tunnel information present' must not be changed into one specifying a tunnel, and the LIR and LIR-pF flags must be passed along unchanged.
Propagating such a route may cause the egress ABR or ASBR to receive Leaf A-D routes that correspond to it, carrying an IP-address-specific Route Target that names the egress ABR or ASBR. It propagates those Leaf A-D routes after changing the Route Target, setting the Global Administrator field to the IP address taken from the S-PMSI A-D route's Next-Hop field or from its Segmented Point-to-Multipoint Next-Hop Extended Community, preferring the Extended Community when it is present.
This lets the ingress PE explicitly track the egress PEs for a given flow even when segmented tunnels are used. However, cross-domain explicit tracking uses S-PMSI A-D routes that do not specify tunnel information, so it can only be done when the S-PMSI A-D route that is a flow's match for tracking is different from the one that is its match for reception.
A tunnel can be split into segments at the borders between areas or autonomous systems, and the question of who wants a flow has to be answered across those boundaries. The rule is that the tracking request has to be passed on unchanged, and the answers that come back have to be re-tagged with the address of the border router that received them.
§6What the Ingress Node Does With the Answer
Consider an ingress node N that receives a Leaf A-D route L, where L carries an IP-address-specific Route Target identifying N, and the Route Key field of L's NLRI is not identical to the NLRI of any current I-PMSI or S-PMSI A-D route originated by N. Under the procedures of RFC 6514 and RFC 7524 such a Leaf A-D route causes no MVPN-specific action to be taken by N.
This document changes that in one case: there is a current wildcard S-PMSI A-D route, originated by N, to which L is a valid response under the procedures of Section 5.2. In that case L must be processed by N.
If L's PTA specifies a tunnel type of Ingress Replication and a non-zero MPLS label, then when N needs to send L a packet belonging to the multicast flow or flows identified in L's NLRI, N must use the specified label. If L's PTA says 'no tunnel information present', or specifies Ingress Replication with an MPLS label of zero, or specifies any other 6514 tunnel type, then what N does with L is a local matter. The Leaf A-D route gives N explicit tracking information for the flow, but that information is for management and monitoring and has no direct effect on the flow of multicast traffic.
Normally an answer that matches no advertisement the ingress node made is ignored. The exception is an answer to a broad request: then it is used. If the answer carries a label, the ingress node must use that label when it sends the traffic; otherwise the answer is used for reporting only.
§8Security Considerations
The security considerations of RFC 6513 and RFC 6514 apply.
By setting the LIR-pF flag in a single wildcard S-PMSI A-D route, a large number of Leaf A-D routes can be elicited. If the flag is set when it is not wanted, through error or malfeasance, a significant increase in control plane overhead can result. Properly protecting the control plane should prevent this kind of attack.
Mitigating such an attack is unfortunately not straightforward. The ingress node can notice that it is receiving, in response to an S-PMSI A-D route whose LIR-pF flag is clear, Leaf A-D routes that have the flag set, and by default the reception of such a route must be logged. Because such log entries can be false positives that generate a lot of noise, implementations should have a knob to disable this logging.
Withdrawing the S-PMSI A-D route would in theory stop the attack, but it is not a good strategy in practice. Under normal operating conditions there are race conditions that may make the ingress node think it is being attacked when it is not. If some egress nodes have a bug that causes them to set the flag when it should be clear, withdrawing the route stops the flow of multicast data to all the egress nodes, causing an unnecessary customer-visible disruption. And the situation that caused the route to be originated in the first place will still exist after the withdrawal, so the route will just be re-originated.
The conclusion is that any action that would ameliorate the effects of this attack would likely have a negative effect during normal operation, so it is better to rely on the security mechanisms that protect the control plane generally.
The new flag can be used to make a router generate a flood of replies. The RFC explains why it does not add a special defence: any targeted response would also break normal operation, so the answer is to protect the control plane in general rather than to add a rule for this one case.