Network Working Group Dino Farinacci
Internet Draft Yakov Rekhter
Expires: June, 1997 cisco Systems
December 1996
Multicast Tag Binding and Distribution using PIM
<draft-farinacci-multicast-tagsw-00.txt>
Status of this Memo
This document is an Internet-Draft. 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."
To learn the current status of any Internet-Draft, please check the
"1id-abstracts.txt" listing contained in the Internet-Drafts Shadow
Directories on ftp.is.co.za (Africa), nic.nordu.net (Europe),
munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or
ftp.isi.edu (US West Coast).
Abstract
This document describes a method for advertising tags for multicast
flows. It strives to use downstream tag assignment to be consistent
with unicast tag distribution. This proposal is media-type
independent. Therefore, it works for multi-access/multicast capable
LANs, point-to-point links, and NBMA networks.
1.0 Overview
We propose to use PIM and combine the (*,G) and (S,G) join state with
tag assignment and distribution. Tags and multicast routes will be
sent together in one message.
1.1 Goals
i. We are motivated to have the upstream Tag Switch Router (TSR) use
Farinacci & Rekhter Multicast Tagging using PIM [Page 1]
Internet Draft November 1996
one tag for multicast data delivery on a network so we can make use
of data-link multicast delivery where available.
ii. We are motivated to use downstream tag assignment to achieve:
o Simplicity and consistency with unicast tag assignment.
o A per interface Tag Information Base (TIB) that guarantees
unique tag assignments on any interface.
o Consistent algorithms for tag assignment and distribution among
different media types.
o Both routing table state and the tag binding information
associated with the state are advertised together in a single
control message thus reducing race conditions.
o Avoid tag reallocation or reassignment when there are RPF
changes (i.e. the multicast distribution tree takes different
shape).
o To improve utilization of tag space by randomizing tag
assignment among all downstream routers joining for a group.
iii. Works with dense-mode or sparse-mode operation.
2.0 Proposal
A TSR that supports multicast sends PIM Join messages on behalf of
hosts that join groups. It sends Joins messages to upstream
neighboring TSRs toward the RP for the shared-tree (*,G) or toward a
source for a source-tree (S,G). If the TSR creates the state for the
group, it will assign a tag for the respective (*,G) or (S,G) state.
It includes the tag in the Join message associated with the multicast
routing table entry. The entry is created in its TIB using the tag as
its incoming tag component.
The upstream TSR, when it receives the Join, will cache the new
multicast routing table state along with the tag. An entry is created
in the TIB and the tag is used as the outgoing component. This tag
will be used by the upstream TSR to forward multicast data packets.
Since PIM Join messages are multicast on a LAN, other downstream
TSRs, that are interested in the group, will hear the message and can
cache the binding of multicast routing table state and tag state
together. Since the upstream TSR is going to forward data packets
using the advertised tag, they must be ready to accept the data
Farinacci & Rekhter Multicast Tagging using PIM [Page 2]
Internet Draft November 1996
packet with that advertised tag.
The first downstream TSR that joins for a group, is the tag assigner
(or called in other forums as the Tag Allocation Server) on a LAN for
a multicast route. All other downstream TSRs that send PIM Join
messages will use the same tag that the assigner selected. A TSR that
sends a PIM Join message with a tag of 0 means that it doesn't know
the tag for the associated multicast routing table entry. When this
occurs, the assigner can trigger a PIM Join message making the tag
known.
This algorithm works on point-to-point links because there is only
one downstream TSR on the link which always becomes the tag assigner.
On NBMA networks, all PIM routers are known to each other through
pseudo-broadcast mechanisms provided by the data-link layer. However,
PIM Join messages are unicast to the upstream TSR. Therefore, other
downstream TSRs will not hear the tag assigner's advertisement. To
overcome this issue, we have each downstream TSR become the tag
assigner on NBMA networks. Since the upstream TSR is going to
pseudo-broadcast the data anyways it can supply a tag for each packet
that goes to each respective downstream TSR.
2.1 Corner cases
Multiple downstream TSRs cannot assign the same tag value for any
multicast route because they partition the tag space into non-
overlapping ranges according to [4]. When a TSR is enabled on an
interface, it obtains a unique tag range for the LAN.
When the tag assigner leaves the group, the tag that it assigned
still remains active. The next highest IP addressed downstream TSR
becomes the owner of that tag and may change it if it sees fit.
However, it is not required to change it. All downstream TSRs can
continue to use the assignment in their Join messages.
If two systems both join for the first time (they do not have state),
at the same time and each choose a different tag value, the highest
IP addressed downstream TSR's tag will be used by the upstream TSR.
The lower addressed TSR will hear the higher addressed TSR's Join too
and will also use it's tag.
If the tag assigner crashes, the highest IP addressed downstream TSR
assigns a new tag to the multicast routes, which were assigned by the
crashing TSR, and triggers a Join message so all other TSRs on the
LAN to use the new tag.
Farinacci & Rekhter Multicast Tagging using PIM [Page 3]
Internet Draft November 1996
When a LAN partitions due to a layer-2 switch failure, it follows the
same logic for the case when a TSR stops joining for a group. When
the partition heals, there may be an RPF neighbor change in one of
the partitions. When there is an RPF neighbor change and the
downstream routers trigger joins to their new RPF neighbor with a
different tag assignment than the other partition is using, one of
two resolutions occur:
1) The TSR which is the allocator in the partition of the new RPF
neighbor will trigger a join if it has a higher IP address than
the allocator in the other region. The downstream routers in the
other partition use the new tag assignment immediately.
2) If the TSR which is the allocator in the partition of the new
RPF neighbor has a lower IP address, all downstream routers and
the new RPF neighbor will switch to the tag assigned by the
allocator in the other partition.
If an RPF change occurs (the topology changed so the upstream TSR is
different), the PIM protocol spec indicates that a PIM Join may be
triggered to get on the new distribution tree as soon as possible. In
this case, if the tag assigner becomes the upstream TSR, then the new
highest IP addressed downstream TSR may become the tag assigner. It
may change the tag if it sees fit. Otherwise, the same tag is used.
3.0 Coexistence of Tag-Capable and Tag-Incapable multicast routers
An upstream router will know if all routers on a subnet are TSRs or
not. If there are any tag incapable routers, the upstream router
will not tag encapsulate multicast data packets. The PIM Hello
message will indicate if the router is tag capable. The PIM Hello
message is sent by every multicast capable router.
If the upstream router detects any non-PIM neighbors on the subnet,
it will assume that they are tag incapable and will not tag
encapsulate multicast data packets.
An optimization may be achieved, if the upstream router knows that
all downstream routers interested in the group are TSRs, it may tag
encapsulate multicast data packets even though there are other tag
incapable routers on the subnet.
Related to the above cases, if there is a group member on a LAN, co-
located with a multicast TSR, only a single packet will be forwarded.
It is the responsibility of the upstream router to decapsulate the
tagged packet and forward it on the LAN as an IP packet so the member
can receive it. The downstream routers may forward the IP packet or
Farinacci & Rekhter Multicast Tagging using PIM [Page 4]
Internet Draft November 1996
tag encapsulate it.
4.0 Tag Conflict Resolution
The use of different data-link layer code-points (i.e. Ethertypes,
PPP protocol types) for unicast and multicast tagswitching allows to
disambiguate between tags associated with unicast routes versus tags
associated with multicast routes. Therefore, the assignment of tags
for unicast routes could be done completely independent from the
assignment of tags for multicast routes, without creating any risk of
ambiguity. For example, the same tag value could be allocated for a
unicast route and for a multicast route.
5.0 Modifications to PIMv2
PIMv2 has a packet format for each address type it may support when
encoding both multicast and unicast addresses. We will define a new
address type called "Tag Address" for unicast address encoding. The
tag will accompany the source address in the Encoded Source Address
format as specified in [2]. The tag value will be in a 32-bit
quantity following the source address. So, for example, an IPv4 Tag
Address format would look like:
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Rsrvd |S|W|R| Mask Len |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Source Address |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Tag |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Refer to [2] for field descriptions.
6.0 Tag Distribution for dense-mode groups
In dense-mode PIM, there is no downstream Join message traveling
upstream to perform the binding of multicast routes with tags.
However, since we don't want a separate algorithm for dense-mode
groups, we extend this basic design for dense-mode PIM.
When a downstream TSR creates (S,G) state from the receipt of 1)
data, or 2) Join/Prune or Graft messages, it will start a periodic
timer to send Join messages with tag assignment information present.
Farinacci & Rekhter Multicast Tagging using PIM [Page 5]
Internet Draft November 1996
The messages look no different and are treated on receipt no
differently than in the sparse-mode case.
The periodic Join message will be multicast on the LAN with an
upstream target address of 0.0.0.0. All multicast TSRs on the LAN
must know the group operates in dense-mode. This is accomplished
using standard PIM mechanisms.
7.0 Security Considerations
Security considerations are not discussed in this memo.
8.0 Acknowledgments
The authors would like to thank Fred Baker and Eric Rosen from cisco
Systems for their insightful comments on this draft.
9.0 Author's Address
Dino Farinacci
Cisco Systems, Inc.
170 Tasman Drive
San Jose, CA, 95134
Email: dino@cisco.com
Yakov Rekhter
Cisco Systems, Inc.
170 Tasman Drive
San Jose, CA, 95134
Email: yakov@cisco.com
10.0 References
[1] Tag Switching Architecture Overview, draft-rfced-tag-switching-
overview-00.txt, Rekhter, Davie, Katz, Rosen, Swallow
[2] Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol
Specification, <draft-ietf-idmr-pim-sm-spec-09.txt>, Estrin,
Farinacci, Helmy, Thaler, Deering, Handley, Jacobson, Liu, Sharma,
Wei, October, 1996
[3] Tag Distribution Protocol, <draft-doolan-tdp-spec-00.txt>,
Doolan, Davie, Katz, Rekhter, Rosen, September, 1996
Farinacci & Rekhter Multicast Tagging using PIM [Page 6]
Internet Draft November 1996
[4] Partitioning Tag Space amoung Multicast Routers on a Common
Subnet, Farinacci, December, 1996
[5] "Tag Switching: Tag Stack Encodings", <draft-rosen-tag-stack-
00.txt>, Rosen, Tappan, Farinacci, Rekhter, Fedorkow, November, 1996
Farinacci & Rekhter Multicast Tagging using PIM [Page 7]