wdiff draft-ietf-idmr-cbt-spec-02.txt draft-ietf-idmr-cbt-spec-03.txt

<draft-ietf-idmr-cbt-spec-02.txt>
Inter-Domain Multicast Routing (IDMR) A. J. Ballardie
INTERNET-DRAFT University College London
N. Jain
Bay Networks, Inc.
S. Reeve
Bay Networks, Inc.

June 20th,

November 21st, 1995

Core Based Trees (CBT) Multicast

-- Protocol Specification --

Status of this Memo

This document is an Internet Draft. Internet Drafts are working do-
cuments of the Internet Engineering Task Force (IETF), its Areas, and
its Working Groups. Note that other groups may also distribute work-
ing documents as Internet Drafts).

Internet Drafts are draft documents valid for a maximum of six
months. Internet Drafts may be updated, replaced, or obsoleted by
other documents at any time. It is not appropriate to use Internet
Drafts as reference material or to cite them other than as a "working
draft" or "work in progress."

Please check the I-D abstract listing contained in each Internet
Draft directory to learn the current status of this or any other
Internet Draft.

Abstract

This document describes the Core Based Tree (CBT) multicast protocol
specification. CBT is a next-generation multicast protocol that makes
use of a shared delivery tree rather than separate per-sender trees
utilized by most other multicast schemes [1, 2, 3].

The

This specification includes a description of an optimization whereby
native IP-style multicasts are forwarded over tree branches as well
as subnetworks with group member presence. This mode of operation
will be called CBT "native mode" and obviates the need to insert a
CBT header into encapsulate
data packets before forwarding over CBT interfaces. Native mode is
only relevant to CBT-only domains or ``clouds''. Also included are
some new "data-driven" features.

A special authors' note is included explaining the primary
differences between this latest specification and the previous
release (June 1995).

The CBT architecture is described in an accompanying document:
draft-ietf-idmr-arch-00.txt. Other related documents include [4, 5].
For all IDMR-related documents, see
http://www.cs.ucl.ac.uk/ietf/idmr.

_1. _D_o_c_u_m_e_n_t _L_a_y_o_u_t

We describe _A_u_t_h_o_r_s' _N_o_t_e

The purpose of this note is to explain how the CBT protocol details by means of example using has
evolved since the topol-
ogy shown previous version (June 1995).

The CBT designers have constantly been seeking to streamline the pro-
tocol and seek new mechanisms to simplify the group initiation pro-
cedure. Especially, it has been a high priority to ensure that the
group joining process is as transparent as possible for new
receivers; ideally, from a user perspective, only a minimum of infor-
mation should be required in figure 1. Examples show how order to join a host joins CBT group -- the
knowledge/input of two group parameters, group address and TTL value,
is a reasonable expectation. At the same time, we strive to keep
join latency to an absolute minimum.

The factor most affecting join latency in CBT is the mechanism by
which each group on a LAN elects a so-called designated router (DR).
This mechanism has now been re-invented, being simpler, and
leaves keeps
join latency to a group, minimum. This new DR election process is explained
in section 2.3.

Core selection, placement, and we also show various tree maintenance scenarios.

In management have prevented a simple
group initiation/joining process, inherent in data-driven schemes
(like DVMRP); some network entity needs to elect a group's cores, and
a mechanism is needed to distribute this figure member hosts information throughout the
network so it is available to potential new receivers.

CBT separates out most aspects of core management from the protocol
itself. This has been made easier due to the fact that core manage-
ment is not a problem unique to CBT, but also PIM-Sparse Mode.
Separate, protocol-independent core management mechanisms are shown as capital letters,
currently being proposed/developed [8, 9]. In the absence of core
management/distribution protocol, the task could be manually handled
by network management facilities.

In CBT, the core routers for a particular group are
prefixed with R, categorised into
PRIMARY CORE, and subnets are prefixed with S.

Figure 1 is shown over...

A B NON-PRIMARY (secondary) CORES.

The core tree, the part of a tree linking all core routers together,
is built on-demand. That is, the core tree is only built subsequent
to a non-primary core receiving a join-request (non-primary core
routers join the primary core router -- the primary need do nothing).
Join-requests carry an ordered list of core routers, making it possi-
ble for the non-primary cores to know where to join.

CBT now supports the aggregation of certain types of control message
on distribution trees, provided aggregation is at all possible. This
depends on coordinated multicast address assignment.

Also catalytic in the simplification of the CBT protocol are the
"multi-protocol support" aspects of the latest proposal of IGMP
(IGMPv3 [6]), in particular, the introduction of the RP/Core-Report
message (see Appendix and [6]).

The end result of these developments is that the CBT protocol is
further simplified and more efficient; six message types have been
eliminated from the previous version of the protocol, thereby reduc-
ing protocol overhead. Furthermore, the new DR election mechanism
ensures group join latency is kept to a minimum.

Throughout this draft, we assume IGMPv3 is operating between hosts
and routers on a LAN.

_2. _P_r_o_t_o_c_o_l _S_p_e_c_i_f_i_c_a_t_i_o_n

_2._1. _C_B_T _G_r_o_u_p _I_n_i_t_i_a_t_i_o_n

A group's initiator elects a small number of candidate cores (which
may be advertised by "some means"). Subsequently, the core distribu-
tion engine (if available) is notified of the new group now associ-
ated with the elected cores. Subsequent network advertisements pro-
vide the <core,group> mapping information for potential new senders
and/or receivers.

_2._2. _T_r_e_e _J_o_i_n_i_n_g _P_r_o_c_e_s_s -- _O_v_e_r_v_i_e_w

It is assumed that hosts receive <core,group> mapping advertisements
via some protocol external to CBT. Given this assumption, the follow-
ing steps are involved in a host joining a CBT tree:

o+ the joining host learns of the candidate cores for the group.

o+ subsequently, an IGMP RP/Core-Report is issued on the subnet-
work, addressed to the corresponding multicast group.

All IGMP messages are received by all operational CBT multicast
routers on the subnetwork. One CBT-capable router per subnetwork
is initially elected as the default LAN CBT DR (DEFAULT DR) for
all groups. This election happens automatically when CBT routers
are initialised. If the subnetwork has multiple CBT routers
present, a (possibly different) group-specific DR (GROUP DR) may
subsequently be elected. This is fully explained in section 2.3.

o+ on receiving an IGMP RP/Core-Report, the local DR takes care of
establishing the subnet as part of the corresponding CBT
delivery tree.

The following CBT control messages come into play during the host
joining process:

o+ JOIN_REQUEST

o+ JOIN_ACK

A join-request is generated by a locally-elected DR (see next sec-
tion) in response to receiving an IGMP group membership report from a
directly connected host. The join is sent to the next-hop on the path
to the target core, as specified in the join packet. The join is pro-
cessed by each such hop on the path to the core, until either the
join reaches the target core itself, or hits a router that is already
part of the corresponding distribution tree (as identified by the
group address). In both cases, the router concerned terminates the
join, and responds with a join-ack, which traverses the reverse-path
of the corresponding join. This is possible due to the transient path
state created by a join traversing a CBT router. The ack simply fixes
that state.

_2._3. _D_R _E_l_e_c_t_i_o_n

Multiple CBT routers may be connected to a multi-access subnetwork.
In such cases it is necessary to elect a (sub)network designated
router (DR) that is responsible for sending IGMP host membership
queries, and for generating join-requests in response to receiving
IGMP group membership reports. Such joins are forwarded upstream by
the DR.

At start-up, a CBT router assumes it is the only CBT-capable router
on its subnetwork. It therefore sends two or three IGMP-HOST-
MEMBERSHIP-QUERYs in short succession (for robustness) in order to
quickly learn about any group memberships on the subnet. If other CBT
routers are present on the same subnet, they will receive these IGMP
queries, and depending on which router was already the elected
querier, yield querier duty to the new router iff the new router is
lower-addressed. If it is not, then the newly-started CBT router will
yield when it hears a query from the already established querier.

The CBT DEFAULT DR (D-DR) is always (exception, next para) the
subnet's IGMP-querier; in CBT these two roles go hand-in-hand. As a
result, there is no protocol overhead whatsoever associated with
electing the CBT D-DR.

On multi-access LANs where different routers may be running different
multicast routing protocols, there may be times when a LAN's
(subnet's) elected querier is a non-CBT router. CBT routers keep
track of their immediate CBT neighbouring routers, and can therefore
easily establish if the source of an IGMP query is CBT-capable or
not. If an elected querier is not CBT-capable, the DR is (implicitly)
elected to be the lowest-addressed neighbour on the same link; if a
CBT router on such a link knows of a lower-addressed neighbour on the
same link, it either does not attempt to claim DR status, or relinqu-
ishes its DR status if it was previously elected DR.

_2._4. _B_a_c_k_w_a_r_d_s _C_o_m_p_a_t_i_b_i_l_i_t_y _w_i_t_h _I_G_M_P_v_1 & _v_2 _H_o_s_t_s

To comply with this specification, CBT routers are expected to run
IGMP version 3 [7]. However, it cannot be assumed that all hosts on a
subnetwork will be running IGMPv3; there may be instances of IGMP
versions 1 and/or 2.

IGMPv1 & v2 hosts will not be able to issue RP/Core Reports,
available with IGMPv3. The implications of this primarily mean that
such hosts must inform a D-DR of <core, group> mappings by means of
network management. Alternatively, hosts may implement minimal user-
level code to emulate IGMPv3-specific messages, and send them as CBT
auxiliary control messages to the specified group address.

NOTE: one recent core distribution proposal [8] does not require
hosts to participate in core election at all. Rather, a local DR
is configured to know a set of core addresses in the lowest level
of a core hierarchy, and a function is used to map a group address
onto a particular core in the hierarchy.

_2._5. _T_r_e_e _J_o_i_n_i_n_g _P_r_o_c_e_s_s -- _D_e_t_a_i_l_s

The receipt of an IGMP group membership report by a CBT D-DR for a
CBT group not previously heard from triggers the tree joining pro-
cess.

Immediately subsequent to receiving an IGMP group membership report
for a CBT group not previously heard from, the D-DR unicasts a JOIN-
REQUEST to the first hop on the (unicast) path to the specified core.
Core information is gleaned either by means of an IGMP RP/Core
Report, also sent in response to an IGMP host membership query, but
prior to an IGMP host membership report, or by some other means.

Each CBT-capable router traversed on the path between the sending DR
and the core processes the join. However, if a join hits a CBT router
that is already on-tree, the join is not propogated further, but
ACK'd from that point.

JOIN-REQUESTs carry the identity of all cores for the group. Assuming
there are no on-tree routers in between, once the join (subcode
ACTIVE_JOIN) reaches the target core, if the target core is not the
primary core (the first listed in the core listing, contained within
the join) it first acknowledges the received join by means of a
JOIN-ACK, then sends a JOIN-REQUEST, subcode REJOIN-ACTIVE, to the
primary core router. Either the primary core, or the first on-tree
router encountered, acknowledges the received rejoin by means of a
JOIN-ACK. Any such router other than the primary core proceeds by
transforming the rejoin into a REJOIN-NACTIVE for loop detection.
This is described in section 6.3.

To facilitate detailed protocol description, we use a sample
topology, illustrated in Figure 1 (shown over). Member hosts are
shown as individual capital letters, routers are prefixed with R, and
subnets are prefixed with S.

A B
| S1 S4 |
------------------- -----------------------------------------------
| | | |
------ ------ ------ ------
| R1 | | R2 | | R5 | | R6 |
------ ------ ------ ------
C | | | | |
| | | | S2 | S8 |
---------- ------------------------------------------ -------------
S3 |
------
| R3 |
| ------ D
| S9 | | S5 |
| | ---------------------------------------------
| |----| | |
---| R7 |-----| ------
| |----| |------------------| R4 |
| S7 | ------ F
| | | S6 |
|-E | ---------------------------------
| |
| ------
|---| |---------------------| R8 |
|R12 -----| ------ G
|---| | | | S10
| S14 ----------------------------
| |
I --| ------
| | R9 |
------
| S12
| ----------------------------
S15 | |
| ------
|----------------------|R10 |
J ---| ------ H
| | |
| ----------------------------
| S13

Figure 1. Example Network Topology

_2. _P_r_o_t_o_c_o_l _S_p_e_c_i_f_i_c_a_t_i_o_n

_2._1. _C_B_T _G_r_o_u_p _I_n_i_t_i_a_t_i_o_n

Like any of
Taking the other multicast schemes, one user, example topology in figure 1, host A is the group initia-
tor, initiates a CBT and has elected core routers R4 (primary core) and R9 (secondary
core) by some external protocol. The <core,group> mapping is subse-
quently advertised by some (possibly same) protocol.

Host A generates an IGMP RP/Core-Report and an IGMP group membership
report when the multicast group. Group initiation could be car-
ried out application is invoked on host A. Both
reports are multicast to the corresponding group address. All multi-
cast routers receive all multicast-addressed messages by default.
The only CBT router on A's subnet (S1) is R1, which is, by default,
the D-DR.

Router R1, receives the RP/Core-Report and the group membership
report, and proceeds to unicast a network management centre, or JOIN-REQUEST, subcode ACTIVE-JOIN
to the next-hop on the path to R4 (R3), the target core in the
RP/Core Report. R3 receives the join, caches the necessary group
information, and forwards it to R4 -- the target of the join.

R4, being the target of the join, sends a JOIN_ACK back out of the
receiving interface to the previous-hop sender of the join, R3. A
JOIN-ACK, like JOIN-REQUESTs, is processed hop-by-hop by some other external
means, rather than have each router
on the reverse-path of the corresponding join. The receipt of a user act as
join-ack establishes the receiving router on the corresponding CBT
tree, i.e. the router becomes part of a branch on the delivery tree.
R3 sends a join-ack to R2, which sends a joinj-ack to R1. A new CBT
branch has been created, attaching subnet S1 to the CBT delivery tree
for the corresponding group.

At this point, it is proposed that IGMP (v3) group initiator. However, in multicasts a
notification across the author's implementation, this flexibility subnet indicating to member hosts that the
delivery tree has been afforded joined successfully. Such a message would
greatly benefit multicast protocols requiring explicit joins [5, 10].

For the
user, and period between any CBT-capable router forwarding (or ori-
ginating) a CBT group is invoked by means of JOIN_REQUEST and receiving a graphical user inter-
face (GUI), known as JOIN_ACK the CBT User Group Management Interface.

NOTE: Work corresponding
router is currently in progress not permitted to address acknowledge any subsequent joins received
for the issue of core
placement.

_2._2. _T_r_e_e _J_o_i_n_i_n_g _P_r_o_c_e_s_s

The following steps are involved in a host establishing itself as
part of a CBT multicast tree:

o+ same group; rather, the joining host must inform all routers on its subnet that router caches such joins till such
time as it
requires has itself received a Designated Router (DR) JOIN_ACK for the group original join. Only
then can it wishes acknowledge any cached joins. A router is said to
join (it be in a
pending-join state if it is awaiting a requirement JOIN_ACK itself.

_2._6. _D-_D_R_s, _G-_D_R_s, _a_n_d _P_r_o_x_y-_a_c_k_s

The DR election mechanism does not guarantee that only one router, the DR, forward
to and from upstream to avoid loops).

o+ the establishment of a DR for the group.

o+ once established, will be the DR must proceed to
router that actually forwards a join off a multi-access network; the distribution
tree.

The following CBT control messages come into play during the host
joining process:

NOTE: all CBT message types are described in section 8 irrespective
of some of
first hop on the comments included with certain message types below.

o+ CORE_NOTIFICATION (sent only by a group initiating host path to
inform each core for the group that it has been elected as a particular core for might be via another
router on the group).

o+ CORE_NOTIFICATION_ACK

o+ DR_SOLICITATION

o+ DR_ADVERTISEMENT_NOTIFICATION (sent only by same (sub)network, which actually forwards off-LAN. It
is not necessary or desirable to have a tree branch rooted anywhere
other than at a local CBT-capable router when that router is unaware of a DR for the group on the
same subnet, interface to and believes it is candidate for from the best next-hop LAN;
only this router off need keep group state information, the LAN to join origina-
tor (D-DR) need not since the core address as specified in first hop is on the
DR_SOLICITATION. This message acts as same LAN. Because
of this, CBT incorporates a tie-breaker in simple mechanism that prevents the case
where there are two or more D-DR
in such routers on a subnet).

o+ DR_ADVERTISEMENT

o+ TAG_REPORT (sent by a joining host to the DR subsequent to
receiving scenarios from keeping group state.

If a DR_ADVERTISEMENT. This message serves join-ack has returned to invoke the
DR to become part originating subnet of the distribution tree, if not already, by
sending a JOIN_REQUEST).

o+ JOIN_REQUEST (sent only by the group's DR iff it is
corresponding join, but has not yet part
of, or in reached the process of, joining originating router of
the corresponding CBT tree).

o+ JOIN_ACK

o+ HOST_JOIN_ACK (multicast across join, obviously the subnet by join-request's first hop is on
the local DR same subnet as an
indication that the DR originating router (the D-DR). A router knows
when it is part of the distribution tree. This
message may be sent in immediate response to receiving a
TAG_REPORT, depending on whether this situation by extracting the DR is already part origin router's subnet
address using its own subnet mask, then comparing the result with its
own address (using address and mask of the
CBT tree or not. subnet that is about to be
forwarded over). If not it one further hop is sent subsequent required for the join-ack to
reach the DR receiv-
ing originator of the corresponding join-request, the router
does not send a JOIN_ACK).

A group-initiating host normal join-ack, but rather sends a CORE-NOTIFICATION message to each JOIN-ACK with
subcode PROXY-ACK. Proxy-acks, like normal join-acks, are unicast.

A router receiving a proxy-ack cancels any transient state it has
created for the corresponding group. The sender of a proxy-ack
becomes the elected cores group-specific DR (G-DR) for the group. This message is acknowledged
(CORE_NOTIFICATION_ACK) by each core individually. Provided at least
one ACK is received group - a host will not be prevented from joining the
tree.

The purpose of token (impli-
cit) identity. In the CORE_NOTIFICATION normal case where there is twofold: firstly, to communi-
cate no LAN extra hop,
the identities of all receipt of a JOIN-ACK means that the cores, together with their rank-
ings, to each of them individually; secondly, to invoke D-DR becomes the building
of G-DR for
the core backbone or core tree. These two procedures follow on one specified group.

Control packets may continue to be incurred an extra-hop if they are
generated by the other in the order just described. New receivers attempting to
join whilst D-DR, but data packets will not; since only the building
sender of the core backbone is still in progress
have their explicit JOIN-REQUEST messages stored by whichever CBT-
capable router involved in proxy-ack keeps a FIB entry for the core joining process is encountered
first.

Taking our example topology in figure 1, host A group, it is the group initia-
tor. The elected cores are
only router R4 (primary core) and R9 (secon-
dary core). Host A first sends a CORE_NOTIFICATION to each on the LAN that has an upstream forwarding entry.

Now let's see an illustration of R4 and
R9, and each responds positively with this; a CORE_NOTIFICATION_ACK.
CORE_NOTIFICATION messages are always unicast.

Subsequent to sending host joins a CORE_NOTIFICATION_ACK, each secondary core
router (in this case there is only one secondary, R9) proceeds CBT group (the
first to
join do so on the primary core, subnet), but more than one router is present on
its subnet. B's subnet, S4, has 3 CBT routers attached. Assume also
that R6 has been elected IGMP-querier and thus forms the core tree, or backbone; R9
unicasts CBT D-DR.

The invoking of a JOIN_REQUEST (subcode CORE_JOIN) multicast application on B causes an IGMP RP/Core-
Report and an IGMP group membership report to R8, its best next-hop be multicast to the primary core, R4. JOIN_REQUESTs (and
corresponding ACKs) are
processed by all intervening CBT-capable routers, and forwarded if
necessary. R8 forwards group. The target core and ordered core list are
contained within the JOIN_REQUEST to RP/Core report. R6 generates a join-request for
target core R4, remembering subcode ACTIVE_JOIN. R6's routing table says the incom-
ing and outgoing interfaces of
next-hop on the JOIN_REQUEST. path to R4 receives the JOIN_REQUEST (subcode CORE_JOIN), realises it is R2, which is on the
target of the join, and therefore sends a JOIN_ACK back out of the
receiving interface same subnet as R6.
This is irrelevant to R6, which unicasts it to R2. R2 unicasts it to
R3, which happens to be already on-tree for the previous-hop sender of the join. R8
receives specified group (from
R1's join). R3 therefore can acknowledge the JOIN_ACK arrived join and forwards unicast
it back to R9 over the interface R2. R2 realises it is not the
join was received from R9. On receipt origin of the JOIN_ACK, R9 need take
no further action. Core tree set up corresponding
join-request, but sees that the origin (R6) is complete.

For on the period between any CBT-capable router forwarding (or ori-
ginating) a JOIN_REQUEST same subnet as
itself, and receiving a JOIN_ACK that over which the corresponding
router is not permitted join-ack would be forwarded to acknowledge any subsequent joins received
for the same group; rather,
origin, R6. R2 unicasts the router caches such joins till such
time as it has itself received a JOIN_ACK for join-ack on its final hop, but sets the original join, at
which time it can acknowledge any cached joins. A router is said
ack subcode to
be PROXY-ACK. This results in a pending-join state if it is awaiting a JOIN_ACK itself.

Returning to host A which has just received both
CORE_NOTIFICATION_ACKs, it must now establish which local CBT router
is DR the D-DR (R6) removing its
pending join information for the specified group. Since A is the group initiator it Another consequence
of receiving a proxy-ack is highly
unlikely that the D-DR need not create a DR FIB entry
for the group will already exist. specified group.

If A was joining an existing group IGMP RP/Core-Report is received by a DR may already be present.

Host A sends D-DR with a DR_SOLICITATION (IP TTL 1) to join for the "all-CBT-routers"
address (224.0.0.7). The solicitation contains one
same group already pending, it takes no action.

Note that the presence of core addresses
as elected underlying transient asymmetric routes is
irrelevant to the tree-building process; CBT tree branches are sym-
metric by the host, to nature in which it wishes they are built. Joins set up transient
state (incoming and outgoing interface state) in all routers along a join
path to be sent. Any
routers on a particular core. The corresponding join-ack traverses the same subnet receiving
reverse-path of the solicitation establish
whether they are join as dictated by the transient state, and not
the path that underlying routing would dictate. Whilst permanent
asymmetric routes could pose a problem for CBT, transient asymmetri-
city is detected by the CBT protocol.

_2._7. _T_r_e_e _T_e_a_r_d_o_w_n

There are two scenarios whereby a tree branch may be torn down:

o+ During a re-configuration. If a router's best next-hop to the
specified core or not. If a
router is one of its existing children, then before
sending the join it must tear down that particular downstream
branch. It does consider itself a candidate and has no record for so by sending a DR
for FLUSH_TREE message which is pro-
cessed hop-by-hop down the group, branch. All routers receiving this
message must process it and forward it multicasts a DR_ADV_NOTIFICATION to all their children.
Routers that have received a flush message will re-establish
themselves on the "all-CBT-
routers" delivery tree if they have directly connected
subnets with group (224.0.0.7). This message acts as presence.

o+ If a tie-breaker in the
case where there is more than one CBT router on the subnet which
thinks has no children it periodically checks all its
directly connected subnets for group member presence. If no
member presence is the best next-hop to the core. The lowest-addressed
source ascertained on any of its subnets it sends a DR_ADV_NOTIFICATION wins the election and subsequently
advertises
QUIT_REQUEST upstream to remove itself as DR by means from the tree.

Let's see, using the example topology of figure 1, how a tree branch
is gracefully torn down using a DR_ADVERTISEMENT, QUIT_REQUEST.

Assume group member B leaves group G on subnet S4. B issues an IGMP
HOST-MEMBERSHIP-LEAVE message which is multicast to the "all-systems "all-routers"
group (224.0.0.1). As R1 is (224.0.0.2). R6, the only router on A's
subnet, it subnet's D-DR and IGMP-querier, responds
with a DR_ADV_NOTIFICATION followed by a
DR_ADVERTISEMENT.

The time between sending a DR_ADV_NOTIFICATION and a DR_ADVERTISEMENT
should be configurable and ideally less than one second group-specific-QUERY. No hosts respond within the required
response interval, so as to keep
join latency to a minimum.

The DR election for subnet S4 D-DR assumes group G traffic is more complex. When host B sends a
DR_SOLICITATION routers R2, R5 and R6 receive it. Assuming R2 and R5
both believe they are the best next-hop to R4 (the specified core)
both send a DR_ADV_NOTIFICATION. no longer
wanted on subnet S4.

Since R2 (the lower addressed) wins the
tie-breaker has no CBT children, and subsequently multicasts a DR_ADVERTISEMENT to S4. All no other directly attached subnets
with joining hosts proceed similarly.

A DR candidate is a router whose outgoing interface, as specified in
its routing table entry for the destination, is different than the
interface over which the DR_SOLICITATION arrived.

On receiving a DR_ADVERTISEMENT host A sends group G presence, it immediately follows on by sending a TAG_REPORT
QUIT_REQUEST to R3, its parent on the DR,
R1. R1 tree for group G. R3 responds
by unicasting a JOIN_REQUEST (subcode ACTIVE_JOIN) to
R3 -- the best next-hop QUIT_ACK to R4, the desired target of the join. R2. R3
forwards (unicast) the received join to R4, remembering incoming subsequently checks whether it in
turn can send a quit by checking group G presence on its directly
attached subnets, and
outgoing interfaces. R4, now already established any group G children. It has the latter (R1 is
its child on tree for the group responds to the JOIN_REQUEST with a JOIN_ACK, G tree), and sends it to
R3, which in turn sends it to R1. The so R3 cannot itself send a quit.
However, the branch R1-R3-R4 is now complete
and part of R3-R2 has been removed from the distribution tree.

On receipt of the JOIN_ACK, R1 multicasts to the "all-systems"
address (224.0.0.1)

_3. _C_B_T _P_r_o_t_o_c_o_l _P_o_r_t_s

CBT routers implement user-level code for tree building, maintenance,
and teardown. This results in a HOST_JOIN_ACK which group-specific forwarding information
base (FIB) being built in user-space. This FIB is a notification downloaded into
kernel-space for fast and efficient data packet forwarding. Any
changes in FIB entries are communicated to the
joining end-system kernel as they occur,
so that the DR has been successful in joining kernel FIB always reflects the current state of any par-
ticular group's tree. The multicast application running on host A can now send data.

Host B proceeds to join

CBT primary and auxiliary control packets then travel inside UDP
datagrams, as the group in a similar fashion, but there following diagram illustrates:

++++++++++++++++++++++++++++++++++++++++++++
| IP header | UDP header | CBT control pkt |
++++++++++++++++++++++++++++++++++++++++++++

Figure 2. Encapsulation for CBT control messages
The following UDP port numbers are
some subtle differences. Host B currently being used (their use at
this stage is not the group initiator unofficial, and it
need not send CORE_NOTIFICATIONs. Host B's first step pending official approval):

o+ CBT Primary control messages - UDP port 7777

o+ CBT Auxiliary control messages - UDP port 7778

_4. _D_a_t_a _P_a_c_k_e_t _F_o_r_w_a_r_d_i_n_g (_n_a_t_i_v_e _m_o_d_e)

In CBT "native mode" only one forwarding method is to elect a
DR, used, namely all
data packets are forwarded over CBT tree interfaces as described above. On receipt of a DR_ADVERTISEMENT from router
R2 in this case, B unicasts a TAG_REPORT to R2. The core specified in
the TAG_REPORT is R4. In response the the TAG_REPORT, R2 unicasts a
JOIN_REQUEST (subcode ACTIVE_JOIN) to R3, native IP mul-
ticasts, i.e. there are no encapsulations required. This assumes that
CBT is the best next-hop to R4. R3
however, has just joined multicast routing protocol in operation within the tree domain
(or "cloud") in question, and so can acknowledge that all routers within the received
join, domain of
operation are CBT-capable, i.e. there are no "tunnels". If this
latter constraint cannot be satisfied it need not travel all the way is necessary to R4. R3 unicasts a
JOIN_ACK encapsulate
IP-over-IP before forwarding to R2, which results in R2 multicasting a HOST_JOIN_ACK
across subnet S4.

_3. child or parent reachable via non-
CBT-capable router(s).

The rules for native mode forwarding are altogether simpler than
those for CBT-mode forwarding (see next section); data packets are
sent over child/parent interfaces as specified in the corresponding
FIB entry, as native IP multicasts. This applies to point-to-point
links as well as broadcast-type subnetworks such as Ethernets.

_5. _D_a_t_a _P_a_c_k_e_t _F_o_r_w_a_r_d_i_n_g (_C_B_T _m_o_d_e)

"CBT mode" as opposed to "native mode" describes the
forwarding/sending forwarding of
data packets over CBT tree interfaces contain-
ing containing a CBT header encapsulation. encap-
sulation. For efficiency, this encapsulation is as follows:

Figure 2. 3. Encapsulation for CBT mode

By using the encapsulations above there is virtually no necessity to modify a
packet's original IP header, and decapsulation header until it is forwarded over subnets with
group member presence in native mode. When this happens, the TTL
value of the original IP header is set to one before forwarding.

The TTL value of the CBT header is set by the encapsulating CBT
router directly attached to the origin of a data packet. This value
is decremented each time it is processed by a CBT router. An encap-
sulated data packet is discarded when the CBT header TTL value
reaches zero.

The purpose of the (outer) encapsulating IP header is to "tunnel"
data packets between CBT-capable routers (or "islands"). The outer IP
header's TTL value is set to the "length" of the corresponding tun-
nel, or MAX_TTL if this is not known, or subject to change.

For native mode IP multicasts, i.e. those without any extra encapsu-
lation, the TTL value of the IP header is decremented each time the
packet is relatively
efficient. received by a multicast router.

It is worth pointing out at this point the distinction between sub-
networks and tree branches, although they can be one and the same.
For example, a multi-access subnetwork containing routers and end-
systems could potentially be both a CBT tree branch and a subnetwork
with group member presence. A tree branch which is not simultaneously
a subnetwork is either a "tunnel" or a point-to-point link.

In CBT forwarding mode there are three forwarding methods used by CBT
routers:

o+ IP multicasting. This method is used to send a data packet
across a directly-connected subnetwork with group member pres-
ence. Thus, system System host changes are not required for CBT. Simi-
larly, Similarly,
end-systems originating multicast data do so in tradi-
tional IP-style. traditional IP-
style.

o+ CBT unicasting. This method is used for sending data packets
encapsulated (as illustrated above) across a tunnel or point-
to-point link. En/de-capsulation takes place in CBT routers.

o+ CBT multicasting. This method sends data packets encapsulated
(as illustrated above) but the outer encapsulating IP header
contains a multicast address. This method is used when a parent
or multiple children are reachable over a single physical inter-
face, as could be the case on a multi-access Ethernet. The IP
module of end-systems subscribed to the same group will discard
these multicasts since the CBT payload type will (protocol id) of the
outer IP header is not be recog-
nized. recognizable by hosts.

CBT routers create Forwarding Information Base (FIB) entries whenever
they send or receive a JOIN_ACK. JOIN_ACK (with the exception of a proxy-ack,
as explained in section 2.5). The FIB describes the parent-child
relationships on a per-group basis. A FIB entry dictates over which
tree interfaces, and how (unicast or multicast) a data packet is to
be sent. Additionally, a data packet is IP multicast over any
directly-connected subnetworks with group member presence. Such
interfaces are kept in a separate table relating to IGMP. A FIB entry
is shown below:

Figure 3. 4. CBT FIB entry

Note that a CBT FIB is required for both CBT-mode and native-mode
multicasting.

The field lengths shown above assume a maximum of 16 directly con-
nected neighbouring routers.

When a data packet arrives at a CBT router, the following rules
apply:

o+ if the packet is an IP-style multicast, it is checked to see if
it originated locally (i.e. if the arrival interface subnetmask
bitwise ANDed with the packet's source IP address equals the
arrival interface's subnet number, the packet was sourced
locally). If
it does not the packet is not of local origin, it is discarded.

o+ the packet is IP multicast to all directly connected subnets
with group member presence. The packet is sent with an IP TTL
value of 1 in this case.

o+ the packet is encapsulated for CBT forwarding (see figure 2) 3) and
unicast to parent and children. However, if more than one child
is reachable over the same interface the packet will be CBT mul-
ticast. Therefore, it is possible that an IP-style multicast and
a CBT multicast will be forwarded over a particular subnetwork.

NOTE: the TTL value of encapsulated data packets is manipulated as
described at the beginning of this section.

Using our example topology in figure 1, let's assume member G ori-
ginates an IP multicast packet. R8 is the DR for subnet S10 (R4 is DR
for all its attached subnets). S10. R8 CBT
unicasts the packet to each of its children, R9 and R12. These children chil-
dren are not reachable over the same interface. R8, being the DR for
subnets S14 and S10 also IP mul-
ticasts multicasts the packet to S14 (S10
received the IP style packet already from the originator). R9, the DR
for S12, need not IP multicast onto S12 since there are no members
present there. R9 CBT unicasts the packet to R10, which is the DR for
S13 and S15. It IP multicasts to both S13 and S15.

Going upstream from R8, R8 CBT unicasts to R4. It is DR for all
directly connected subnets and therefore IP multicasts the data
packet onto S5, S6 and S7, all of which have member presence. R4 uni-
casts the packet to all outgoing children, R3 and R7 (NOTE: R4 does
not have a parent since it is the primary core router for the group).
R7 IP multicasts onto S9. R3 CBT unicasts to R1 and R2, its children.
Finally, R1 IP multicasts onto S1 and S3, and R2 IP multicasts onto
S4.

_3._1.

_5._1. _N_o_n-_M_e_m_b_e_r _S_e_n_d_i_n_g (_C_B_T _m_o_d_e)

For a multicast data packet to span beyond the scope of the originat-
ing subnetwork at least one CBT-capable router must be present on
that subnetwork. The default DR (D-DR) for the group on the
subnetwork must encap-
sulate encapsulate the IP-style packet and unicast it to a
core for the group. This requires CBT routers to have access to a
mapping mechanism between group addresses and core routers. This
mechanism is currently beyond the scope of this document.

_4. _D_a_t_a _P_a_c_k_e_t _F_o_r_w_a_r_d_i_n_g (_n_a_t_i_v_e _m_o_d_e)

Alternatively, hosts could perform the CBT "native mode" only one forwarding method encapsulation themselves,
but this would require hosts to run a core discovery protocol. Host
modifications required for such a protocol, and the subsequent data
packet encapsulation, are considered extremely undesirable, and are
therefore not considered further.

_5._2. _E_l_i_m_i_n_a_t_i_n_g _t_h_e _T_o_p_o_l_o_g_y-_D_i_s_c_o_v_e_r_y _P_r_o_t_o_c_o_l _i_n _t_h_e _P_r_e_s_e_n_c_e _o_f
_T_u_n_n_e_l_s

Traditionally, multicast protocols operating within a virtual topol-
ogy, i.e. an overlay of the physical topology, have required the
assistance of a multicast topology discovery protocol, such as that
present in DVMRP. However, it is used, namely possible to have a multicast proto-
col operate within a virtual topology without the need for a multi-
cast topology discovery protocol. One way to achieve this is by hav-
ing a router configure all
data packets its tunnels to its virtual neighbours in
advance. A tunnel is identified by a local interface address and a
remote interface address. Routing is replaced by "ranking" each such
tunnel interface associated with a particular core address; if the
highest-ranked route is unavailable (tunnel end-points are forwarded over required
to run an Hello-like protocol between themselves) then the next-
highest ranked available route is selected, and so on.

CBT trees are built using the same join/join-ack mechanisms as
before, only now some branches of a delivery tree interfaces run in native mode,
whilst others (tunnels) run in CBT mode. Underlying unicast routing
dictates which interface a packet should be forwarded over. Each
interface is configured as either native mode or CBT mode, so a
packet can be encapsulated (decapsulated) accordingly.

As an example, router R's configuration would be as follows:

intf type mode remote addr
-----------------------------------
#1 phys native IP mul-
ticasts, i.e. there are no encapsulations required. This assumes that -
#2 tunnel cbt 128.16.8.117
#3 phys native -
#4 tunnel cbt 128.16.6.8
#5 tunnel cbt 128.96.41.1

core backup-intfs
--------------------
A #5, #2
B #3, #5
C #2, #4

The CBT FIB needs to be slightly modified to accommodate an extra
field, "backup-intfs" (backup interfaces). The entry in this field
specifies a backup interface whenever a tunnel interface specified in
the FIB is down. Additional backups (should the multicast routing protocol first-listed backup
be down) are specified for each core in operation within the domain
(or "cloud") in question. It also assumes that all routers within core backup table. For
example, if interface (tunnel) #2 were down, and the
domain target core of operation are CBT-capable, i.e. there are no "tunnels". If
this latter constraint cannot be satisfied it is necessary to encap-
sulate IP-over-IP before forwarding to a child or parent reachable
via non-CBT-capable router(s).

Besides the structural characteristics of "native mode" data packets,
described above, the data
CBT control packet forwarding rules are identical were core A, the core backup table suggests using
interface #5 as a replacement. If interface #5 happened to
those described in section 3.

_4._1. be down
also, then the same table recommends interface #2 as a backup for
core A.

_5._3. _N_o_n-_M_e_m_b_e_r _S_e_n_d_i_n_g (_n_a_t_i_v_e _m_o_d_e)

For a multicast data packet to span beyond the scope of the originat-
ing subnetwork at least one CBT-capable router must be present on
that subnetwork. The default DR for the group (D-DR) on the subnetwork must encap-
sulate (IP-over-IP) the IP-style packet and unicast it to a core for
the group. This requires CBT routers to have access to a mapping
mechanism between group addresses and core routers. This mechanism
is currently beyond the scope of this document.

_5.

Again, host changes could obviate the need for a local router to per-
form a <core, group> mapping and an encapsulation, but this is not
considered a desirable option.

_6. _T_r_e_e _M_a_i_n_t_e_n_a_n_c_e

Once a tree branch has been created, i.e. a CBT router has received a
JOIN_ACK for a JOIN_REQUEST previously sent (forwarded), a child
router is required to monitor the status of its parent/parent link at
fixed intervals by means of a ``keepalive'' mechanism operating
between them. The ``keepalive'' mechanism is implemented by means of
two CBT control messages: CBT_ECHO_REQUEST and CBT_ECHO_REPLY.
Immediately subsequent to a parent/child relationship being esta-
blished, a child unicasts a CBT-ECHO-REQUEST to its parent, which
unicasts a CBT-ECHO-REPLY in response.

CBT echo requests and replies may be aggregated to conserve bandwidth
on links over which tree branches overlap. However, this is only pos-
sible if group address assignment has been coordinated to facilitate
aggregation. (see section 8.4).

For any non-core CBT router, if its parent router, or path to the parent,
fails, that non-core router the child is initially responsible for re-attaching itself,
and therefore all routers subordinate to it on the same branch, to
the tree.

_5._1.

_6._1. _R_o_u_t_e_r _F_a_i_l_u_r_e

A non-core

An on-tree router can detect a failure from the following two cases:

o+ if a child stops receiving CBT_ECHO_REPLY messages. In this case
the child realises that its parent has become unreachable and
must therefore try and re-connect to the tree. It does so by The router on the
tree immediately subordinate to the failed router arbitrarily choosing an alternate
elects a core from its list of cores for this group. It establishes a chosen core's reachability by
unicasting a CBT_CORE_PING message to it, to which the core
responds with a CBT_PING_REPLY. On receipt of the latter, the
re-joining The rejoin-
ing router then sends a JOIN_REQUEST (subcode ACTIVE_REJOIN) ACTIVE_JOIN if it
has no children attached, and subcode ACTIVE_REJOIN if at least
one child is attached) to the best next-hop router on the path
to the elected core. A router
will continue arbitrarily choosing If no JOIN-ACK is received after the speci-
fied number of retransmissions, an alternate core is arbitarily
elected from the core list. The process is repeated until a
CBT_PING_REPLY
JOIN-ACK is received for a maximum of RECONNECT-TIMEOUT seconds
(90 secs is received. the recommended default).

o+ if a parent stops receiving CBT_ECHO_REQUESTs from a child. In
this case the parent simply removes the child interface from its
FIB entry for the particular group.

_5._2.

_6._2. _R_o_u_t_e_r _R_e-_S_t_a_r_t_s

There are two cases to consider here:

o+ Core re-start. In this case, the core router relies on receiving
a CBT_CORE_PING message, which contains All JOIN-REQUESTs (all types) carry the list identi-
ties (i.e. addresses) of each of the cores for
the specified a group. Obviously, one of the If a
router is a core addresses for a group, but has only recently re-started,
it will not be its own. If aware that it is a core realises its core status for any group(s). In such
circumstances, a group core only becomes aware that it is such by
receiving a JOIN-REQUEST. Subsequent to a core learning its
status in this way, if it is not the primary core it ack-
nowledges the received join, then sends a JOIN_REQUEST (sub-
code ACTIVE_JOIN) (subcode
ACTIVE_REJOIN) to the primary core. If the re-started router in ques-
tion is
the primary core, it need not send a join, but rather awaits
joins and considers itself part of take no action, i.e. in all cir-
cumstances, the tree again. primary core simply waits to be joined by other
routers.

o+ Non-core re-start. In this case, the router can only join the
tree again if a downstream router sends a JOIN_REQUEST through
it, or it is elected DR for one of its directly attached sub-
nets.

_5._3.
nets, and subsequently receives an IGMP RP/Core Report.

_6._3. _R_o_u_t_e _L_o_o_p_s

Routing loops are only a concern when a router with at least one
child is attempting to re-join a CBT tree. In this case the re-
joining router sends a JOIN_REQUEST (subcode ACTIVE REJOIN) to the
best next-hop on the path to the core. This join is forwarded as nor-
mal until it reaches either the core core, or a non-core router that is
already part of the tree. If the join reaches the specified core, the
join terminates there and is ACKd as normal. If however, the join is
terminated by non-core router, the ACTIVE_REJOIN is converted to a
NON_ACTIVE_REJOIN
NON_ACTIVE_REJOIN, keeping the origin as that specified in the
ACTIVE_REJOIN, and forwarded upstream. A JOIN_ACK is also sent
downstream down-
stream to acknowledge the received join.

The NON_ACTIVE_REJOIN is a loop detection packet. All routers receiving receiv-
ing this must forward it over their parent interface. If This process
continues until the NON_ACTIVE_REJOIN is received by the primary core
for the group, or the NON_ACTIVE_REJOIN is received by the originator
of the corresponding ACTIVE_REJOIN. A router will know this since the
"origin" field remains unchanged when a join is converted from an
ACTIVE_REJOIN to a NON_ACTIVE_REJOIN. In the former case, the
primary core acknowledges the NON_ACTIVE_REJOIN with JOIN-ACK, sub-
code NACTIVE_REJOIN. This message is unicast directly to the originator of
REJOIN_ACTIVE originator. In the correspond-
ing ACTIVE_REJOIN should receive latter case, the NON_ACTIVE_REJOIN it ACTIVE_REJOIN ori-
ginator immediately sends a QUIT_REQUEST to its recently established newly-established
parent and the loop is broken.

o+ Using figure 4 5 (over) to demonstrate this, if R3 is attempting
to re-join the tree (R1 is the core in figure 4) 5) and R3 believes
its best next-hop to R1 is R6, and R6 believes R5 is its best
next-hop to R1, which sees R4 as its best next-hop to R1 -- a
loop is formed. R3 begins by sending a JOIN_REQUEST (subcode
ACTIVE_REJOIN, since R4 is its child) to R6. R6 forwards the
join to R5. R5 is on-tree for the group, so changes the join
subcode to NON_ACTIVE_REJOIN, and forwards this to its parent,
R4. R4 forwards the NON_ACTIVE_REJOIN to R3, its parent. R3
originated the corresponding ACTIVE_REJOIN, and so it immedi-
ately sends a QUIT_REQUEST to R6, which in turn sends a quit if
it has not received an ACK from R5 already AND has itself a
child or subnets with member presence. If so it need does not send a
quit -- the loop has been broken by R3 sending the first quit.

QUIT_REQUESTs are typically acknowledged by means of a QUIT_ACK, but
there might be cases where, due to failure, the parent cannot
respond. In this case the child nevertheless removes the parent
information after some small number (typically 3) of re-tries.

------
| R1 |
------
|
---------------------------
|
------
| R2 |
------
|
---------------------------
| |
------ |
| R3 |--------------------------|
------ |
| |
--------------------------- |
| | ------
------ | | |
| R4 | |-------| R6 |
------ | |----|
| |
--------------------------- |
| |
------ |
| R5 |--------------------------|
------ |
|

Figure 4: 5: Example Loop Topology

_6. _D_a_t_a _P_a_c_k_e_t _L_o_o_p_s

NOTE: this is only applicable when CBT header encapsulation

In the other scenario where no loop is in
use.

When a data packet hits its first on-tree router, that actually formed, router is
responsible for setting R3
sends a join, subcode REJOIN_ACTIVE to R2, the on-tree bits in next-hop on the CBT header. This
indicates path
to all subsequent routers on core R1. R2 forwards the tree that re-join to R1, the packet is in primary core, which
unicasts a JOIN-ACK to the process originator of spanning the tree for REJOIN_ACTIVE, i.e. the group. However, it might be
that a misbehaving router forwards an on-tree packet over a non-tree
interface, and such
join-ack remains invisible to R2.

_7. _D_a_t_a _P_a_c_k_e_t _L_o_o_p_s

The CBT protocol builds a packet might work its way back onto the tree,
potentially forming loop-free distribution tree. If all routers
that comprise a particular tree function correctly, data packet loop. Therefore, the on-tree bits
in the CBT header serve to identify such packets --
should never traverse a router
receive tree branch more than once.

CBT routers will only forward native-style data packets if they are
received over a valid on-tree interface. A native-style data packet with its on-tree bits set
that is not received over a non-tree such an interface the packet is immediately discarded.

_7. _T_r_e_e _T_e_a_r_d_o_w_n

There are two scenarios whereby a tree branch may be torn down:

o+ During a re-configuration, if

Encapsulated CBT data packets from a router's best next-hop to the
specified core non-member sender can arrive via
an "off-tree" interface (this is one of its existing children then before send-
ing the re-join it must tear down that particular downstream
branch. It does so by sending how CBT-mode sends data across tun-
nels, and how data from non-member senders in native-mode or CBT-mode
reaches a FLUSH_TREE message tree). The encapsulating CBT data packet header includes
an "on-tree" field, which is pro-
cessed hop-by-hop down contains the branch. All routers receiving value 0x00 until the data
packet reaches an on-tree router. At this
message point, the router must process it and forward it con-
vert this value to all their children.
Routers that have received a flush message will re-establish
themselves on the delivery tree if they have directly connected
subnets with group presence. Subsequent 0xff to sending a FLUSH_TREE, indicate the router can send the re-join to its child.

o+ If a CBT router has no children it periodically checks all its
directly connected subnets for group member presence. If no
member presence data packet is ascertained on any of its subnets it sends a
QUIT_REQUEST upstream to remove itself now on-tree.
This value remains unchanged, and from here on the tree.

With regards packet should
traverse only on-tree interfaces. If an encapsulated packet happens
to "wander" off-tree and back on again, the latter scenario, lets see using the example
topology of figure 1 how a tree branch is torn down.

Assume member E leaves on-tree router
will receive the group (if IGMPv2 is in use CBT encapsulated packet via an explicit
IGMP_LEAVE message off-tree interface.
However, this router will be sent by E). If R7 registers no further
group presence (by means recognise that the "on-tree" field of IGMP) then R7 sends a QUIT_REQUEST to R4.
R4 responds with a QUIT_ACK the
encapsulating CBT header is set to R7. R4 has children AND subnets with
group presence, 0xff, and so does not itself attempt to quit immediately discards
the tree. The
branch R4-R7 has been torn down. packet.

_8. _C_B_T _P_a_c_k_e_t _F_o_r_m_a_t_s _a_n_d _M_e_s_s_a_g_e _T_y_p_e_s

CBT packets travel in IP datagrams. We distinguish between two types
of CBT packet: CBT data packets, and CBT control packets.

CBT data packets carry a CBT header when these packets are traversing
CBT tree branches. The enscapsulation (for "CBT mode") is shown
below:

Figure 5. 6. Encapsulation for CBT mode
CBT control packets carry a CBT control header. All CBT control mes-
sages are implemented over UDP. This makes sense for several reasons:
firstly, all the information required to build a CBT delivery tree is
kept in user space. Secondly, implementation is made considerably
easier.

CBT control messages fall into two categories: primary maintenance
messages, which are concerned with tree-building, re-configuration,
and teardown, and auxiliary maintenance messsages, which are mainly
concerned with general tree maintenance.

_8._1. _C_B_T _H_e_a_d_e_r _F_o_r_m_a_t

See over....

Figure 6. 7. CBT Header

Each of the fields is described below:

o+ Vers: Version number -- this release specifies version 1.

o+ type: indicates whether the payload is data or control infor-
mation.

o+ hdr length: length of the header, for purpose of checksum
calculation.

o+ protocol: upper-layer protocol number. on-tree: indicates whether the packet is on-tree (0xff) or
off-tree (0x00). Once this field is set (i.e. on-tree), it
is non-changing.

o+ checksum: the 16-bit one's complement of the one's complement
of the CBT header, calculated across all fields.

o+ IP TTL: TTL value gleaned from the IP header where the packet
originated. It is decremented each time it traverses a CBT
router.

o+ on-tree: indicates whether the packet is on- or off-tree.
Once this field is set (i.e. on-tree), it is non-changing.

o+ group identifier: multicast group address.

o+ core address: the unicast address of a core for the group. A
core address is always inserted into the CBT header by an
originating host, since at any instant, it does not know if
the local DR for the group is on-tree. If it is not, the
local DR must unicast the packet to the specified core.

o+ packet origin: source address of the originating end-system.

o+ flow-identifier: (T.B.D) value uniquely identifying a previously previ-
ously set up data stream.

o+ security fields: these fields (T.B.D.) will ensure the
authenticity and integrity of the received packet.

_8._2. _C_o_n_t_r_o_l _P_a_c_k_e_t _H_e_a_d_e_r _F_o_r_m_a_t

See over...

The individual fields are described below. It should be noted that the
contents of the only
certain fields beyond ``group identifier'' are empty in some processed for the dif-
ferent control messages: messages.

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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| vers |unused | type | code | unused # cores |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| hdr length | checksum |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| group identifier |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| packet origin |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| target core address |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Core #1 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Core #2 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Core #3 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Core #4 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Core #5 .... |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Resource Reservation fields |
| (T.B.D) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| security fields |
| (T.B.D) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 7. 8. CBT Control Packet Header

o+ Vers: Version number -- this release specifies version 1.

o+ type: indicates control message type (see sections 1.3, 1.4).

o+ code: indicates sub-code subcode of control message type.

o+ # cores: number of core addresses carried by this control
packet.

o+ header length: length of the header, for purpose of checksum
calculation.

o+ checksum: the 16-bit one's complement of the one's complement
of the CBT control header, calculated across all fields.

o+ group identifier: multicast group address.

o+ packet origin: source address of the originating end-system.

o+ target core address: desired/actual core affiliation of control mes-
sage. con-
trol message.

o+ Core #Z: Maximum of 5 core addresses may be specified for any
one group. An implementation is not expected to utilize more
than, say, 3.

NOTE: It was an engineering design decision to have a fixed max-
imum number IP address of core addresses, to avoid a variable-sized packet. #Z.

o+ Resource Reservation fields: these fields (T.B.D.) are used
to reserve resources as part of the CBT tree set up pro-
cedure.

o+ Security fields: these fields (T.B.D.) ensure the authenti-
city and integrity of the received packet.

_8._3. _P_r_i_m_a_r_y _M_a_i_n_t_e_n_a_n_c_e _M_e_s_s_a_g_e _T_y_p_e_s

There are six types of CBT primary maintenance message, namely: message. Primary mes-
sage subcodes are described in the next section.

o+ JOIN-REQUEST: invoked by an end-system, JOIN-REQUEST (type 1): generated and sent
(unicast) by a CBT router and unicast to
the specified core address. It is processed hop-by-hop on its
way to the specified core. Its purpose is to establish the
sending CBT router, and all intermediate CBT routers, as part
of the corresponding delivery tree.

o+ JOIN-ACK: JOIN-ACK (type 2): an acknowledgement to the above. The full
list of core addresses is carried in a JOIN-ACK, together
with the actual core affiliation (the join may have been terminated ter-
minated by an on-tree router on its journey to the specified
core, and the terminating router may or may not be affiliated
to the core specified in the original join). A JOIN-ACK
traverses the same path as the corresponding JOIN-REQUEST, and it
with each CBT router on the path processing the ack. It is
the receipt of a JOIN-ACK that actually creates a tree
branch.

o+ JOIN-NACK: JOIN-NACK (type 3): a negative acknowledgement, indicating
that the tree join process has not been successful.

o+ QUIT-REQUEST: QUIT-REQUEST (type 4): a request, sent from a child to a
parent, to be removed as a child to that parent.

o+ QUIT-ACK: QUIT-ACK (type 5): acknowledgement to the above. If the
parent, or the path to it is down, no acknowledgement will be
received within the timeout period. This results in the
child nevertheless removing its parent information.

o+ FLUSH-TREE: FLUSH-TREE (type 6): a message sent from parent to all children, chil-
dren, which traverses a complete branch. This message results
in all tree interface information being removed from each
router on the branch, possibly because of a re-configuration
scenario.

_8._3._1. _P_r_i_m_a_r_y _M_a_i_n_t_e_n_a_n_c_e _M_e_s_s_a_g_e _S_u_b_c_o_d_e_s

The JOIN-REQUEST has three valid sub-codes, namely JOIN-ACTIVE, RE-
JOIN-ACTIVE, and RE-JOIN-NACTIVE.

A JOIN-ACTIVE is subcodes:

o+ ACTIVE-JOIN (code 0) - sent from a CBT router that has no
children for the specified group.

A RE-JOIN-ACTIVE is

o+ REJOIN-ACTIVE (code 1) - sent from a CBT router that has at
least one child for the specified group.

A RE-JOIN-NACTIVE originally started out as an active re-join, but
has reached an on-tree router for the corresponding group. At this
point, the router changes the join status to non-active re-join and
forwards it on its parent branch, as does each CBT router that
receives it. Should the router that originated the active re-join
subsequently receive the non-active re-join, it must immediately send

o+ REJOIN-NACTIVE (code 2) - converted from a QUIT-REQUEST to its parent router. It then attempts to re-join
again. In this way REJOIN-ACTIVE by
the re-join acts as a loop-detection packet.

_8._4. _A_u_x_i_l_l_i_a_r_y _M_a_i_n_t_e_n_a_n_c_e _M_e_s_s_a_g_e _T_y_p_e_s

There are eleven CBT auxilliary maintenance message types:

o+ CBT-DR-SOLICITATION: first on-tree router receiving a request sent from REJOIN-ACTIVE. This mes-
sage is forwarded over a host to router's parent interface until it
either reaches the CBT
``all-routers'' multicast address, for primary core, or is received by the address origi-
nator of the
best next-hop CBT corresponding REJOIN-ACTIVE.

A JOIN-ACK has three valid subcodes:

o+ NORMAL (code 0) - sent by a core router, or on-tree non-core
router on the LAN acknowledging joins with subcodes REJOIN-ACTIVE and
ACTIVE-JOIN.

o+ PROXY-ACK (code 1) - acknowledgement of a join-request by a
router connected to the core same subnet as specified
in the solicitation. originator (subnet
D-DR) of the corresponding join.

o+ CBT-DR-ADVERTISEMENT: REJOIN-NACTIVE (code 2) - sent by a reply primary core to ack-
nowledge the above. Advertisements
are addressed receipt of a join-request received with subcode
REJOIN-NACTIVE. This ack is unicast directly to the ``all-systems'' multicast group.

o+ CBT-CORE-NOTIFICATION: unicast from a group initiating host router
that converted the corresponding REJOIN-ACTIVE to each core selected for REJOIN-
NACTIVE. The CBT control packet "origin" field contains the group, this message notifies
each core
IP address of the identities of each originator of the other core(s) REJOIN-ACTIVE, so in
order for the group, together with their primary core ranking. The receipt to directly reach the source of
this message invokes
the building REJOIN-NACTIVE, the converting router inserts its IP
address in the "core address" field of the control packet
header. The primary core tree by all
cores other than uses the highest-ranked (primary core).

o+ CBT-CORE-NOTIFICATION-ACK: a notification of acceptance address in this field to
becoming a core for
determine the target of the join-ack, subcode REJOIN-NACTIVE.

_8._4. _A_u_x_i_l_l_i_a_r_y _M_a_i_n_t_e_n_a_n_c_e _M_e_s_s_a_g_e _T_y_p_e_s

There are two CBT auxilliary maintenance message types. CBT auxiliary
messages are encoded in a group, to CBT control packet header, and the corresponding end-system. fields
of the control packet are interpreted as illustrated below. The
interpretation of certain fields further depends on whether aggrega-
tion and security are implemented.

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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| vers |unused | type | code | aggregate |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| hdr length | checksum |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| group identifier (or low end of range) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| group id mask or NULL |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| NULL (if security implemented) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| security fields if implemented or NULL |
| (T.B.D) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 9. CBT Echo Request/Reply
o+ CBT-ECHO-REQUEST: CBT-ECHO-REQUEST (type 7): once a tree branch is established,
this messsage acts as a ``keepalive'', and is unicast from
child to parent.

o+ CBT-ECHO-REPLY: CBT-ECHO-REPLY (type 8): positive reply to the above.

o+ CBT-CORE-PING: unicast from a

CBT router to a core when a
tree router's parent has failed. The purpose of this message Echo Requests/Replies can be sent as aggregates, or individually
for each group if multicast address assignment is to establish core reachability before sending a JOIN-
REQUEST to it.

o+ CBT-PING-REPLY: positive reply to the above.

o+ CBT-TAG-REPORT: unicast from an end-system to such that aggrega-
tion is not possible. If aggregation is implemented, the designated
router for "aggregate"
field (which replaces the corresponding group, subsequent to "# cores" field of the end-
system receiving a designated router advertisement (as well
as a core notification reply if group-initiating host). This
message invokes standard control
packet header. In this case, no cores are assumed present in the sending of a JOIN-REQUEST if mes-
sage) will contain the receiv-
ing router value 0xff, otherwise 0x00.

If aggregation is not already part of the corresponding tree.

o+ CBT-HOST_JOIN_ACK: group-specific multicast by a CBT router
that originated a JOIN-REQUEST on behalf of some end-system
on implemented, the same LAN (subnet). The purpose of this message "group id mask" field is set
to
notify end-systems NULL, or is not present, depending on the LAN belonging whether security is imple-
mented or not. Masks are used according to the specified
group their standard networking
usage.

The "flow-id" field (to be done) of such things as: success in joining the delivery
tree; actual core affiliation.

o+ CBT-DR-ADV-NOTIFICATION: multicast to the CBT ``all-routers''
address, this message standard control packet
header is sent subsequent to receiving a CBT-
DR-SOLICITATION, but prior NULL if security is implemented, not present otherwise.

The security fields (to be done) are only present if security is
implemented.

_9. _D_e_f_a_u_l_t _T_i_m_e_r _V_a_l_u_e_s

There are several CBT control messages which are transmitted at fixed
intervals. These values, retransmission times, and timeout values,
are given below. Note these are recommended default values only, and
are configurable with each implementation (all times are in seconds):

o+ CBT-ECHO-INTERVAL 30 (time between sending successive CBT-ECHO-
REQUESTs to parent).

o+ PEND-JOIN-INTERVAL 10 (retransmission time for join-request if
no ack rec'd)

o+ PEND-JOIN-TIMEOUT 30 (time to any CBT-DR-ADVERTISEMENT being
sent. It acts as try joining a tie-breaking mechanism should more than
one router on the subnet think itself the best next-hop to
the addressed core. It also promts an already established DR
to announce itself as such if it different core, or
give up)

o+ EXPIRE-PENDING-JOIN 90 (remove transient state for join that has
not already done so in
response been ack'd)
o+ CBT-ECHO-TIMEOUT 90 (time to a CBT-DR-SOLICITATION.

_9. consider parent unreachable)

o+ CHILD-ASSERT-INTERVAL 90 (check last time we rec'd an ECHO from
each child)

o+ CHILD-ASSERT-EXPIRE-TIME 180 (remove child information if no
ECHO received)

o+ IFF-SCAN-INTERVAL 300 (scan all interfaces for group presence.
If none, send QUIT)

_1_0. _I_n_t_e_r_o_p_e_r_a_b_i_l_i_t_y _I_s_s_u_e_s

One of the design goals of CBT is for it to fully interwork with
other IP multicast schemes. We have already described how CBT-style
packets are transformed into IP-style multicasts, and vice-versa.

In order for CBT to fully interwork with other schemes, it is neces-
sary to define the interface(s) between a ``CBT cloud'' and the cloud
of another scheme. The CBT authors are currently working out the
details of the ``CBT-other'' interface, and therefore we omit further
discussion of this topic at the present time.

_1_0.

_1_1. _C_B_T _S_e_c_u_r_i_t_y _A_r_c_h_i_t_e_c_t_u_r_e

see current I-D: draft-ietf-idmr-mkd-02.txt draft-ietf-idmr-mkd-01.{ps,txt}

Acknowledgements

Special thanks goes to Paul Francis, NTT Japan, for the original
brainstorming sessions that brought about this work.

Thanks also to the networking team at Bay Networks for their comments
and sugges-
tions, suggestions, in particular Steve Ostrowski for his suggestion of
using "native mode" as a router optimization, Eric Crawley, Scott
Reeve, and Nitin Jain. Thanks also to Ken Carlberg (SAIC) for review-
ing the text, and generally providing constructive comments
throughout.

I would also like to thank the participants of the IETF IDMR working
group meetings for their general constructive comments and sugges-
tions since the inception of CBT.

APPENDIX

IGMP version 3 has recently been proposed [6]. The authors have the
following recommendations for amendments (all minor) to IGMPv3:

o+ The IGMPv3 draft [6] introduces a new IGMP message type, the PIM
RP-REPORT message. Its message format is shown below:

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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Code | Checksum |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Group Address |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Version | Reserved | # of RP's (N) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| RP Address [1] |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| RP Address [...] |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| RP Address [N] |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 10. PIM RP-REPORT.

The CBT authors propose the following minor amendments to the IGMP
PIM RP-REPORT:

o+ the report to be re-named RP/CORE-REPORT

o+ RP fields re-named RP/Core fields

o+ the reserved field to be re-named the "target core" field, to
contain the numeric value of the position of the target core in
the RP/Core list

o+ The introduction of a new code value to distinguish PIM RP
reports from CBT Core reports.

These minor amendments to IGMPv3 would satisfy CBT's operational
requirements.

Author's Address:

Tony Ballardie,
Department of Computer Science,
University College London,
Gower Street,
London, WC1E 6BT,
ENGLAND, U.K.

Tel: ++44 (0)71 419 3462
e-mail: A.Ballardie@cs.ucl.ac.uk

Nitin Jain,
Bay Networks, Inc.
3 Federal Street,
Billerica, MA 01821,
USA.

Tel: ++1 508 670 8888
e-mail: njain@BayNetworks.com

Scott Reeve,
Bay Networks, Inc.
3 Federal Street,
Billerica, MA 01821,
USA.

Tel: ++1 508 670 8888
e-mail: sreeve@BayNetworks.com

References

[1] DVMRP. Described in "Multicast Routing in a Datagram Internet-
work", S. Deering, PhD Thesis, 1990. Available via anonymous ftp from:
gregorio.stanford.edu:vmtp/sd-thesis.ps.

[2] J. Moy. Multicast Routing Extensions to OSPF. Communications of
the ACM, 37(8): 61-66, August 1994.

[3] D. Farinacci, S. Deering, D. Estrin, and V. Jacobson. Protocol
Independent Multicast (PIM) Dense-Mode Specification (draft-ietf-
idmr-pim-spec-01.ps). Working draft, 1994.

[4] A. J. Ballardie. Scalable Multicast Key Distribution (draft-ietf-
idmr-mkd-02.txt).
idmr-mkd-01.txt). Working draft, 1995.

[5] A. J. Ballardie. "A New Approach to Multicast Communication in a
Datagram Internetwork", PhD Thesis, 1995. Available via anonymous ftp
from: cs.ucl.ac.uk:darpa/IDMR/ballardie-thesis.ps.Z.

[6] W. Fenner. Internet Group Management Protocol, version 2 (IGMPv2),
(draft-idmr-igmp-v2-01.txt).

[7] B. Cain, S. Deering, A. Thyagarajan. Internet Group Management
Protocol Version 3 (IGMPv3) (draft-cain-igmp-00.txt).

[8] M. Handley, J. Crowcroft, I. Wakeman. Hierarchical Rendezvous
Point proposal, work in progress.
(http://www.cs.ucl.ac.uk/staff/M.Handley/hpim.ps).

[9] D. Estrin et al. USC/ISI, Work in progress. (document not yet
available).

[10] D. Estrin et al. PIM Sparse Mode Specification. (draft-ietf-
idmr-pim-sparse-spec-00.txt).