When building datagram shortest-path trees in the presence of
areas, it is often the case that the source of the datagram and
(at least some of) the destination group members are in separate
areas. Since detailed topological information concerning one
**FROM**
|RT|RT|RT|RT|RT|RT|
|1 |2 |3 |4 |5 |7 |N3|
----- -------------------
RT1| | | | | | |0 |
RT2| | | | | | |0 |
RT3| | | | | | |0 |
* RT4| | | | | | |0 |
* RT5| | |14|8 | | | |
T RT7| | |20|14| | | |
O N1|3 | | | | | | |
* N2| |3 | | | | | |
* N3|1 |1 |1 |1 | | | |
N4| | |2 | | | | |
Ia,Ib| | |15|22| | | |
N6| | |16|15| | | |
N7| | |20|19| | | |
N8| | |18|18| | | |
N9-N11,H1| | |19|16| | | |
N12| | | | |8 |2 | |
N13| | | | |8 | | |
N14| | | | |8 | | |
N15| | | | | |9 | |
Figure 6: Area 1's MOSPF database.
Networks and routers are represented by vertices.
An edge of cost X connects Vertex A to Vertex B iff
the intersection of Column A and Row B is marked
with an X. In addition, RT1, RT2 and N3 are labelled
with multicast group A, RT1 is labelled with multicast
group B, and both RT3 and RT4 are labelled as
wild-card multicast receivers.
**FROM**
|RT|RT|RT|RT|RT|RT|RT
|3 |4 |5 |6 |7 |10|11|
------------------------
RT3| | | |6 | | | |
RT4| | |8 | | | | |
RT5| |8 | |6 |6 | | |
RT6|8 | |7 | | |5 | |
RT7| | |6 | | | | |
* RT10| | | |7 | | |2 |
* RT11| | | | | |3 | |
T N1|4 |4 | | | | | |
O N2|4 |4 | | | | | |
* N3|1 |1 | | | | | |
* N4|2 |3 | | | | | |
Ia| | | | | |5 | |
Ib| | | |7 | | | |
N6| | | | |1 |1 |3 |
N7| | | | |5 |5 |7 |
N8| | | | |4 |3 |2 |
N9-N11,H1| | | | | | |1 |
N12| | |8 | |2 | | |
N13| | |8 | | | | |
N14| | |8 | | | | |
N15| | | | |9 | | |
Figure 7: The backbone's MOSPF database.
Networks and routers are represented by vertices.
An edge of cost X connects Vertex A to Vertex B iff
the intersection of Column A and Row B is marked
with an X. In addition, RT3 and RT4 are labelled
with both multicast groups A and B, and RT7, RT10,
and RT11 are labelled with multicast group A.
OSPF area is not distributed to other OSPF areas (the flooding
of router-LSAs, network-LSAs and group-membership-LSAs is
restricted to a single OSPF area only), the building of complete
datagram shortest-path trees is often impossible in the inter-
area case. To compensate, approximations are made through the
use of wild-card multicast receivers and OSPF summary-link-LSAs.
When it first receives a datagram for a particular [source net,
destination group] pair, a router calculates a separate datagram
shortest-path tree for each of the router's attached areas. Each
datagram shortest-path tree is built solely from LSAs belonging
to the particular area's link state database. Suppose that a
router is calculating a datagram shortest-path tree for Area A.
It is useful then to separate out two cases.
The first case, Case 1: The source of the datagram belongs to
Area A has already been described in Section 2.3.2. However, in
the presence of OSPF areas, during tree pruning care must be
taken so that the branches leading to other areas remain, since
it is unknown whether there are group members in these (remote)
areas. For this reason, only those branches having no group
members nor wild-card multicast receivers are pruned when
producing the datagram shortest-path tree.
As an example, suppose in Figure 4 that Host H2 sends a
multicast datagram to destination Group A. Then the datagram's
shortest-path tree for Area 1, built identically by all routers
in Area 1 that receive the datagram, is shown in Figure 8. Note
that both inter-area multicast forwarders (RT3 and RT4) are on
the datagram's shortest-path tree, ensuring the delivery of the
datagram to the backbone and from there to Areas 2 and 3.
o Case 2: The source of the datagram belongs to an area other
than Area A. In this case, when building the datagram
shortest-path tree for Area A, the immediate neighborhood of
the datagram's source is unknown. However, there are
summary-link-LSAs in the Area A link state database
indicating the cost of the paths between each of Area A's
inter-area multicast forwarders and the datagram source.
These summary links are used to approximate the neighborhood
of the datagram's source; the tree begins with links
directly connecting the source to each of the inter-area
multicast forwarders. These links point in the reverse
o RT3 (W, origin=N4)
|
1|
|
N3 (Mb) o
/ \
0/ \0
/ \
RT2 (Ma,Mb) o o RT4 (W)
Figure 8: Datagram's shortest-path tree,
Area 1, source N4, destination Group A
direction (towards instead of away from the datagram source)
from the links considered in Case 1 above. All additional
links added to the tree also point in the reverse direction.
The final datagram shortest-path tree is then produced by,
as before, pruning all branches having no group-members nor
wild-card multicast receivers.
As an example, suppose again that Host H2 in Figure 4 sends
a multicast datagram to destination Group A. The datagram's
shortest-path tree for the backbone is shown in Figure 9.
The neighborhood around the source (Network N4) has been
approximated by the summary links advertised by routers RT3
and RT4. Note that all links in Figure 9's datagram
shortest-path tree have arrows pointing in the reverse
direction, towards Network N4 instead of away from it.
The reverse costs used for the entire tree in Case 2 are forced
because summary-link-LSAs only specify the cost towards the
datagram source. In the presence of asymmetric link costs, this
may lead to less efficient routes when forwarding multicasts
o N4
/ \
2/ \3
/ \
RT3 (Ma,Mb) o o RT4 (Ma,Mb)
/ \
6/ \8
/ \
RT6 o o RT5
| |
5| |6
| |
RT10 (Ma) o o RT7 (Ma)
|
2|
|
RT11 (Ma) o
Figure 9: Datagram shortest-path tree: Backbone,
source N4, destination Group A. Note that
reverse costs (i.e., toward origin) are
used throughout.
between areas.
Those routers attached to multiple areas must calculate multiple
trees and then merge them into a single forwarding cache entry.
As shown in Section 2.3.2, when connected to a single area the
router's position on the datagram shortest-path tree determines
(in large part) its forwarding cache entry. When attached to
multiple areas, and hence calculating multiple datagram
shortest-path trees, each tree contributes to the forwarding
cache entry's list of downstream interfaces/neighbors. However,
only one of the areas' datagram shortest-path trees will
determine the forwarding cache entry's upstream node. When one
of the attached areas contains the datagram source, that area
will determine the upstream node. Otherwise, the tiebreaking
rules of Section 12.2.7 are invoked.
Consider again the example of Host H2 in Figure 4 sending a
multicast datagram to destination Group A. Router RT3 will
calculate two datagram shortest-path trees, one for Area 1 and
one for the backbone. Since the source of the datagram (Host
H2) belongs to Area 1, the Area 1 datagram shortest-path tree
determines RT3's upstream node: Network N4. Router RT3
calculates two downstream interfaces for the datagram: the
interface to Network N3 (which comes from Area 1's datagram
shortest-path tree) and the serial line to Router RT6 (which
comes from the backbone's datagram shortest-path tree). As for
Router RT10, it calculates two trees, determining its upstream
node from the backbone tree and its two downstream interfaces
from the Area 2 tree. Finally, Router RT11 calculates three
trees, determining its upstream node from the Area 2 tree and
its downstream interface from the Area 3 tree.
4. Inter-AS multicasting
This section explains how MOSPF deals with the forwarding of
multicast datagrams between Autonomous Systems. Certain AS boundary
routers in a MOSPF system will be configured as inter-AS multicast
forwarders. It is assumed that these routers will also be running an
inter-AS multicast routing protocol. This specification does not
dictate the operation of such an inter-AS multicast routing
protocol. However, the following interactions between MOSPF and the
inter-AS routing protocol are assumed:
(1) MOSPF guarantees that the inter-AS multicast forwarders will
receive all multicast datagrams; but it is up to each router so
designated to determine whether the datagram should be forwarded
to other Autonomous Systems. This determination will probably be
made via the inter-AS routing protocol.
(2) MOSPF assumes that the inter-AS routing protocol is forwarding
multicast datagrams in an RPF (reverse path forwarding; see
[Deering] for an explanation of this terminology) fashion. In
other words, it is assumed that a multicast datagram whose
source (call it X) lies outside the MOSPF domain will enter the
MOSPF domain at those points that are advertising (into OSPF)
the best routes back to X. MOSPF calculates the path of the
datagram through the MOSPF domain based on this assumption.
MOSPF designates an inter-AS multicast forwarder as a wild-card
multicast receiver in all of its attached areas. As in the inter-
area case, this ensures that the routers remain on all pruned
shortest-path trees and thereby receive all multicast datagrams,
regardless of destination.
As an example, suppose that in Figure 1 both RT5 and RT7 were
configured as inter-AS multicast forwarders. Then the link state
database would look like the one pictured in Figure 2, with the
addition of a) wild-card status for RT5 and RT7 (they would appear
with superscripts of "w") and b) the external links originated by
RT5 and RT7 being labelled as multicast-capable[12].
As another example, consider the area configuration in Figure 4.
Again suppose RT5 and RT7 are configured as inter-AS multicast
forwarders. Then in Area 1's link state database (Figure 6), the
external links originated by RT5 and RT7 would again be labelled as
multicast-capable. However, note that in Area 1's database RT5 and
RT7 are not labelled as wild-card multicast receivers. This is
unnecessary; since Area 1's inter-area multicast forwarders (RT3 and
RT4) are wild-cards, all multicast datagrams will be forwarded to
the backbone. And in the backbone's link state database (Figure 7),
RT5 and RT7 will be labelled as wild-cards.
4.1. Building inter-AS datagram shortest-path trees.
When multicast datagrams are to be forwarded between Autonomous
Systems, the datagram shortest-path tree is built as follows.
Remember that the router builds a separate tree for each area to
which it is attached; these trees are then merged into a single
forwarding cache entry. Suppose that the router is building the
tree for Area A. We break up the tree building into three cases.
This first two cases have already been described earlier in this
memo: Case 1 (the source of the datagram belongs to Area A)
having been described in Section 2.3.2 and Case 2 (the source of
the datagram belongs to another OSPF area) having been described
in Section 3.2. The only modification to these cases is that
inter-AS multicast forwarders, as well as group members and
inter-area multicast forwarders, must remain on the pruned
trees. The new case is as follows:
o Case 3: The source of the datagram belongs to another
Autonomous System. The immediate neighborhood of the source
is then unknown. In this case the multicast-capable AS
external links are used to approximate the neighborhood of
the source; the tree begins with links directly attaching
the source to one or more inter-AS multicast forwarders. The
approximating AS external links point in the reverse
direction (i.e., towards the source), just as with the
approximating summary links in Case 2. Also, as in Case 2,
all links included in the tree must point in the reverse
direction. The final datagram shortest-path tree is then
produced (as always) by pruning those branches having no
group members nor wild-card multicast receivers.
As an example, suppose that a host on Network N12 (see
Figure 4) originates a multicast datagram for Destination
Group B. Assume that all external costs pictured are OSPF
external type 1 metrics. Then any routers in Area 1
receiving the datagram would build the datagram shortest-
path tree pictured in Figure 10. Note that all links in the
tree point in the reverse direction, towards the source. The
tree indicates that the routers expect the datagram to enter
the Autonomous System at Router RT7, and then to enter the
area at Router RT4.
Note that in those cases where the "best" inter-AS multicast
forwarder is not directly attached to the area, the
neighborhood of the source is actually approximated by the
concatenation of a summary link and a multicast-capable AS
external link. This is in fact the case in Figure 10.
In Case 3 (datagram source in another AS) the requirement that
all tree links point in the reverse direction (towards the
source) accommodates the fact that summary links and AS external
links already point in the reverse direction. This also leads to
the requirement that the inter-AS multicast routing protocol
operate in a reverse path forwarding fashion (see condition 2 of
Section 4). Note that Reverse path forwarding can lead to sub-
optimal routing when costs are configured asymmetrically. And it
can even lead to non-delivery of multicast datagrams in the case
of asymmetric reachability.
Inter-AS multicast forwarders may end up calculating a
forwarding cache entry's upstream node as being external to the
AS. As an example, Router RT7 in Figure 10 will end up
calculating an external router (via its external link to Network
o N12
|
2|
|
o RT7
|
14|
|
o RT4 (W)
|
0|
|
o N3 (Mb)
/|\
/ | \
1/ | 1\
/ 1| \
/ | \
RT1 (Mb) o | o RT3 (W)
o
RT2 (Ma,Mb)
Figure 10: Datagram shortest-path tree: Area 1,
source N12, destination Group B. Note that
reverse costs (i.e., toward origin) are
used throughout.
N12) as the upstream node for the datagram. This means that RT7
must receive the datagram from a router in another AS before
injecting the datagram into the MOSPF system.
4.2. Stub area behavior
AS external links are not imported into stub areas. Suppose that
the source of a particular datagram lies outside of the
Autonomous System, and that the datagram is forwarded into a
stub area. In the stub area's datagram shortest-path tree the
neighborhood of the datagram's source cannot be approximated by
AS external links. Instead the neighborhood of the source is
approximated by the default summary links (see Section 3.6 of
[OSPF]) that are originated by the stub area's intra-area
multicast forwarders.
Except for this small change to the construction of a stub
area's datagram shortest-path trees, all other MOSPF algorithms
(e.g., merging with other areas' datagram shortest-path trees to
form the forwarding cache) function the same for stub areas as
they do for non-stub areas.
4.3. Inter-AS multicasting in a core Autonomous System
It may be the case that the MOSPF routing domain connects
together many different Autonomous Systems, thereby serving as a
"core Autonomous System" (e.g, the old NSFNet backbone). In this
case, it could very well be that the majority of the MOSPF
routers are also inter-AS multicast forwarders. Having each
inter-AS multicast forwarder then declare itself a wild-card
multicast receiver could very well waste considerable network
bandwidth. However, as an alternative to declaring themselves
wild-card multicast receivers, the inter-AS multicast routers
could instead explicitly advertise all groups that they were
interested in forwarding (to other "client" Autonomous Systems)
in group-membership-LSAs. These advertised groups would have to
be learned through an inter-AS multicast routing protocol (or
possibly even statically configured).
This in essence allows the clients of the core Autonomous System
to advertise their group membership into the core. However,
since any client MOSPF domains will still have their inter-AS
multicast forwarders configured as wild-card multicast
receivers, this advertisement will be asymmetric: the core will
not advertise its or others' group membership to the clients.
The achieves the same inter-AS multicast routing architecture
that MOSPF uses for inter-area multicast routing (see Figure 5).
5. Modelling internal group membership
A MOSPF router may itself contain multicast applications. A typical
example of this is a UNIX workstation that doubles as a multicast
router. This section concerns two alternative ways of representing
the group membership of the MOSPF router's internal applications.
Both representations have advantages. For maximum flexibility, the
MOSPF forwarding algorithm (see Section 11) has been specified so
that either representation can be used in a MOSPF router (and in
fact, both representations can be used at once, depending on the
application).
The first representation is based on the paradigm presented in RFC
1112. In this case, an application joins a multicast group on one or
more specific physical interfaces. The application then receives a
multicast datagram if and only if it is received on one of the
specified interfaces. If a datagram is received on multiple
specified interfaces, the application receives multiple copies.
Figure 11 shows this algorithm as it is implemented in (modified)
BSD UNIX kernels. The figure shows the processing of a multicast
datagram, starting with its reception on a particular interface.
First copies of the datagram are given to those applications that
have joined on the receiving interface. Then the forwarding decision
(pictured as a box containing a question mark) is made, and the
packet is (possibly) forwarded out certain interfaces. If these
interfaces are not capable of receiving their own multicasts, a copy
of the datagram must be internally looped back to appropriately
joined applications.
The advantages to the RFC1112 representation are as follows:
o It is the standard for the way an IP host joins multicast
groups. It is simplest to use the same membership model for
hosts and routers; most would consider an IP router to be a
special case of an IP host anyway.
o It is the way group membership has been implemented in BSD UNIX.
Existing multicast applications are written to join multicast
groups on specific interfaces.
o The possibility of receiving multiple datagram copies may
improve fault tolerance. If the datagram is dropped due to an
+-------+
|receive|
+-------+
|
|---> To application
|
+-------------------+
|forwarding decision|
+-------------------+
|
/ \
/---\----> To application
/ \------> To application
/ \
/ \
+--------+ +--------+
|transmit| |transmit|
+--------+ +--------+
Figure 11: RFC1112 representation of internal
group membership
error on the path to some interface, another interface may still
receive a copy.
o The ability to specify a particular receiving interface may
improve the accuracy of IP multicast's expanding ring search
mechanism (see Section 2.3.4).
o Membership in the non-routable multicast groups (224.0.0.1 -
224.0.0.255) must be on a per-interface basis. An OSPF router
always belongs to 224.0.0.5 (AllSPFRouters) on its OSPF
interfaces, and may belong to 224.0.0.6 (AllDRouters) on one or
more of its OSPF interfaces.
The second representation is MOSPF-specific. In this case, an
application joins a multicast group on an interface-independent
basis. In other words, group membership is associated with the
router as a whole, not separately on each interface. The application
then receives a copy of a multicast datagram if and only if the
datagram would actually be forwarded by the MOSPF router. Figure 12
shows how this algorithm would be implemented. The datagram is
received on a particular interface. If the datagram is validated for
forwarding (i.e., the receiving interface connects to the matching
forwarding cache entry's upstream node), a copy of the datagram is
also given to appropriately joined applications. Note that this
model of group membership is not as general as the RFC1112 model,
in that it can only be implemented in MOSPF routers and not in
arbitrary IP hosts. However, it has the following advantages:
o The application does not need to have knowledge of the router
interfaces. It does not need to know what kind or how many
interfaces there are; this will be taken care of by the MOSPF
protocol itself.
o As long as any interface is operational, the application will
continue to receive multicast datagrams. This happens
automatically, without the application modifying its group
membership.
o The application receives only one copy of the datagram. Using
the RFC1112 representation, whenever an application joins on
more than one interface (which must be done if the application
does not want to rely on a single interface), multiple datagram
copies will be received during normal operation.
6. Additional capabilities
This section describes the MOSPF configuration options that allow
routers of differing capabilities to be mixed together in the same
+-------+
|receive|
+-------+
|
|
|
+-------------------+
|forwarding decision|---> to application
+-------------------+
|
/ \
/ \
/ \
/ \
/ \
+--------+ +--------+
|transmit| |transmit|
+--------+ +--------+
Figure 12: MOSPF-specific representation of internal
group membership
routing domain. Note that these options handle special circumstances
that may not be encountered in normal operation. Default values for
the configuration settings are specified in Appendix B.
6.1. Mixing with non-multicast routers
MOSPF routers can be mixed freely with routers that are running
only the base OSPF algorithm (called non-multicast routers in
the following). This allows MOSPF to be deployed in a piecemeal
fashion, thereby speeding deployment and allowing
experimentation with multicast routing on a limited scale.
When a MOSPF router builds a datagram shortest-path tree, it
omits all non-multicast routers. For example, in Figure 1, if
Router RT6 was not a multicast router, the datagram shortest-
path tree in Figure 3 would be built with a more circuitous
branch through Router RT5, instead of through Router RT6. In
addition, non-multicast routers do not participate in the
flooding of the new group-membership-LSAs. This adheres to the
general principle that a router should not have to handle those
link state advertisements whose format (or contents) the router
does not understand.
Mixing MOSPF routers with non-multicast routers creates a number
of potential problems. Certain mixings of MOSPF and non-
multicast routers can cause multicast datagrams to take
suboptimal paths, or in other cases can lead to the non-delivery
of multicast datagrams. In addition, mixing MOSPF routers and
non-multicast routers can cause the paths of multicast datagrams
to diverge radically from the path of unicast datagrams. Such
divergences can make routing problems harder to debug.
In particular, the following specific difficulties may arise
when mixing MOSPF routers with non-multicast routers:
o Even though there is unicast connectivity to a destination,
there may not be multicast connectivity. For example, if
Router RT10 in Figure 1 becomes a non-multicast router, the
group member connected to Network N11 will no longer be able
to receive multicasts sourced by Host H2. But the two hosts
will be able to exchange unicasts (e.g., ICMP pings).
o When the Designated Router for a multi-access network is a
non-multicast router, the network will not be used for
forwarding multicast datagrams. For example, if in Figure 1
Router RT4 is Designated Router for Network N3, and RT4 is
non-multicast, Network N3 will not be used to forward IP
multicasts. This would mean that multicast datagrams
originated by Hosts H2 and H3 would not be forwarded beyond
their local network (N4), even though it seems that the
needed multicast connectivity exists.
o When forwarding multicast datagrams between areas, mixing of
MOSPF routers and non-multicast routers in the source area
may cause unexpected loss of multicast connectivity. This is
because in the inter-area routing of multicast datagrams the
neighborhood of the datagram's source is approximated by
OSPF summary links, and OSPF summary-link-LSAs do not carry
indications/guarantees of the summarized path's multicast
routing capability.
6.2. TOS-based multicast
MOSPF allows a separate datagram shortest-path tree to be built
for each IP Type of Service. This means that the path of a
multicast datagram can vary depending on the datagram's TOS
classification, as well as its source and destination.
For each router interface, OSPF allows a separate metric to be
configured for each IP TOS. When building the shortest path tree
for TOS X, the cost of a path is the sum of the component
interfaces' TOS X metrics. Note that OSPF requires that a TOS 0
metric be specified for each interface. However, as a form of
data compression, metrics need only be specified for non-zero
TOS if they are different than the TOS 0 metric.
Additionally, OSPF routers can be configured to ignore TOS when
forwarding packets. Such routers, called TOS-incapable, build
only the TOS 0 portion of the routing table. TOS-incapable
routers can be mixed freely with TOS-capable routers when
forwarding unicast packets. The way this is handled for unicast
packets is that the unicast is forwarded along the TOS 0 route
whenever the TOS X route does not exist. However, MOSPF must
treat this situation somewhat differently, since each router
must build the exact same tree rooted at the datagram's source.
Like OSPF, MOSPF allows TOS-based routing to be optional. TOS-
capable and TOS-incapable multicast routers can be mixed freely
in the routing domain. TOS-incapable routers will only ever
build TOS 0 datagram shortest-path trees. TOS-capable routers
will first build TOS 0 datagram shortest-path trees. If these
trees contain only TOS-capable routers, datagram shortest-path
trees are then built separately for non-zero TOS values.
Otherwise, the TOS 0 datagram shortest-path tree is used to
forward all traffic, regardless of its TOS designation. Using
this logic, all routers in essence continue to utilize identical
datagram shortest-path trees. See Section 12.2.8 for more
details.
6.3. Assigning multiple IP networks to a physical network
Assigning multiple IP networks/subnets to a single physical
network causes some confusion in MOSPF. This is because the
underlying OSPF protocol treats these IP networks/subnets as
entirely separate entities, originating separate network-LSAs
for each and forming separate adjacencies for each, while IGMP
recognizes only the single underlying physical network. Adding
to the problem is the fact that when a multicast datagram is
received from such a multiply-addressed physical wire, there is
no good way to choose the datagram's upstream node (which must
be done in order to make the forwarding decision; see Section 11
for details). As a result, unless this situation is dealt with
through configuration, unwanted replication of multicast
datagrams may occur when they are forwarded over multiply-
addressed wires.
As a remedy, MOSPF allows multicast forwarding to be disabled on
certain IP networks/subnets. When multicast forwarding is
disabled on the wire's "extra" subnets (i.e., all but one), the
extra subnets will not appear in datagram shortest-path trees,
nor will they appear in local group database or forwarding cache
entries. As a result, the possibility of unwanted datagram
replication is eliminated. The actual disabling of multicast
forwarding on a subnet is done through setting the
IPMulticastForwarding parameter to disabled on all router
interfaces connecting to the subnet (see Section B.2).
6.4. Networks on Autonomous System boundaries
Another complication can arise on IP networks/subnets that lie
on the boundary of a MOSPF Autonomous System. Similar to the
unicast situation where these networks may be running multiple
IGPs (Interior Gateway Protocols), these networks may also be
running multiple multicast routing protocols. It may then become
impossible for a MOSPF router to determine whether a multicast
datagram is being forwarded along the datagram shortest-path
tree, or whether it has been inadvertently received from the
other Autonomous System. Guessing wrong can lead to either
unwanted replication or non-delivery of the multicast datagram.
In addition, in order to prevent receiving duplicate multicast
datagrams, group members on these boundary networks will
probably want to declare their membership to one Autonomous
System and not another.
For example, consider the two Autonomous Systems pictured in
Figure 13. Network X is on the boundary of both ASes. One
possible multicast datagram path is shown; the datagram
originates in a third Autonomous System, and is then delivered
to both AS #1 and AS #2 separately. The paths through the two
Autonomous Systems may end up having certain boundary networks
as common segments. In Figure 13, Network X is common to both
paths. In this case, if both Autonomous Systems were running
(separate copies of) MOSPF, the same datagram would appear twice
on Network X as a data-link multicast. This would cause
duplicate datagrams to be received by any group members on
Network X or downstream from Network X.
MOSPF has two mechanisms to eliminate this replication of
multicast datagrams. First, a system administrator can configure
certain networks to forward multicast datagrams as data-link
unicasts instead of data-link multicasts. This is done by
setting the IPMulticastForwarding parameter to data-link unicast
on those router interfaces attaching to the network (see Section
B.2). As an example, in Figure 13 the routers in AS #2 could be
configured so that Router C would send the multicast datagram
out onto Network X as a data-link unicast addressed directly to
Router D. Router D would accept this data-link unicast, but
<-Datagram path->*
* *
* *
* .....*.........
.........*..... | . * AS #2
AS #1 * . |*****+---+
+---+*****|*----|RTC|
|RTA|----*|* . +---+
+---+ . *|* .
. *|* .
. *|* . +---+
+---+ . *|*----|RTD|
|RTB|----*|*****+---+
+---+*****| .....*..........
.........*.... | *
* | *
* Network X *
*
Figure 13: Networks on AS boundaries
would reject any data-link multicast forwarded by Router A. This
would eliminate replication of multicast datagrams downstream
from Network X. In addition, if the IPMulticastForwarding
parameter is set to data-link unicast on Network X, group
membership will not be monitored on the network. This will
prevent group members attached directly to Network X from
receiving multiple datagram copies, since group membership on
the boundary network will be monitored from only one AS (AS #1
in our example).
It should be noted that forwarding IP multicasts as data-link
unicasts has some disadvantages when three or more MOSPF routers
are attached to the network. First of all, it is more work for a
router to send multiple unicasts than a single multicast.
Second, the multiple unicasts consume more network bandwidth
than a single multicast. And last, it increases the delay for
some group members since multiple unicasts also take longer to
send than a single multicast.
6.5. Recommended system configuration
In order to make MOSPF's selection of routes more predictable,
it is recommended that all routers in any particular OSPF area
have the same multicast and TOS capabilities.Keeping areas
homogeneous ensures that IP multicast packets will follow
relatively the same path as IP unicasts. In contrast, while
heterogeneous areas will function, and will probably be
necessary at least during the initial introduction of multicast
routing, such areas may produce seemingly sub-optimal and
unexpected routes. For example, see Section 6.1 above for a
detailed description of the possible pitfalls when mixing
multicast and non-multicast routers.
As for the other options presented above, to achieve the most
predictable results it is recommended that a router interface's
IPMulticastForwarding parameter be set to a value other than
data-link multicast only when either a) multiple IP networks
have been assigned to a single physical wire or b) multiple
multicast routing protocols are running on the attached network.
7. Basic implementation requirements
An implementation of MOSPF requires the following pieces of system
support. Note that this support is in addition to that required for
the base OSPF implementation as outlined in Section 4.4 of [OSPF].
o Promiscuous multicast reception. In a multicast router, it is
necessary to receive all IP multicasts at the data-link level.
On those interfaces where IP multicast datagrams are
encapsulated by a wide range of data-link multicast destination
addresses (e.g, ethernet and FDDI), this is most easily
accomplished by disabling any hardware filtering of multicast
destinations (i.e., by "opening up" the interface's multicast
filter).
o Data-link multicast/broadcast detection. To avoid unwanted
replication of multicast datagrams in certain exceptional
conditions, it is necessary for the multicast router to
determine whether a datagram was received as a data-link
multicast/broadcast or as a data-link unicast, for later use by
the MOSPF forwarding mechanism. See Section 6.4 for more
details.
o An implementation of IGMP. MOSPF uses the Internet Group
Management Protocol (IGMP, documented in [RFC1112]) to monitor
multicast group membership. See Section 9 for details.
8. Protocol data structures
The MOSPF protocol is described herein in terms of its operation on
various protocol data structures. These data structures are included
for explanatory uses only, and are not intended to constrain a MOSPF
implementation. Besides the data structures listed below, this
specification will also reference the various data structures (e.g.,
OSPF interfaces and neighbors) defined in [OSPF].
In a MOSPF router, the following items are added to the list of
global OSPF data structures described in Section 5 of [OSPF]:
o Local group database. This database describes the group
membership on all attached networks for which the router is
either Designated Router or Backup Designated Router. This in
turn determines the group-membership-LSAs that the router will
originate, and the local delivery of multicast datagrams (see
Sections 2.3.1 and 10).
o Forwarding cache. Each entry in the forwarding cache describes
the path of a multicast datagram having a particular [source
net, multicast destination, TOS] combination. These cache
entries are calculated when building the datagram shortest-path
trees. See Sections 2.3.4 and 11 for more details.
o Multicast routing capability. Indicates whether the router is
running the multicast extensions defined in this memo. A router
running the multicast extensions must still run the base OSPF
algorithm as set forth in [OSPF]. Such a router will continue to
interoperate with non-multicast-capable OSPF routers when
forwarding IP unicast traffic.
o Inter-area multicast forwarder. Indicates whether the router
will forward IP multicasts from one OSPF area to another. Such a
router declares itself a wild-card multicast receiver in its
non-backbone area router-LSAs (see Section 14.6), and also
summarizes its attached areas' group membership to the backbone
in group-membership-LSAs. When building inter-area datagram
shortest-path trees, it is these routers that appear immediately
adjacent to the datagram source at the root of the tree (see
Section 3.2). Not all multicast-capable area border routers need
be configured as inter-area multicast forwarders. However,
whenever both ends of a virtual link are multicast-capable, they
must both be configured as inter-area multicast forwarders (see
Section 14.11).
o Inter-AS multicast forwarder. Indicates whether the router will
forward IP multicasts between Autonomous Systems. Such a router
declares itself a wild-card multicast receiver in its router-
LSAs (see Section 14.6). These routers are also assumed to be
running some kind of inter-AS multicast protocol. They mark all
external routes that they import into the OSPF domain as to
whether they provide multicast connectivity (see Section 14.9).
When building inter-AS multicast datagram trees, it is these
routers that appear immediately adjacent to the datagram source
at the root of the tree.
8.1. Additions to the OSPF area structure
The OSPF area data structure is described in Section 6 of
[OSPF]. In a MOSPF router, the following item is added to the
OSPF area structure:
o List of group-membership-LSAs. These link state
advertisements describe the location of the area's multicast
group members. Group-membership-LSAs are flooded throughout
a single area only. Area border routers also summarize their
attached areas' membership by originating group-membership-
LSAs into the backbone area. For more information, see
Sections 3.1 and 10.
8.2. Additions to the OSPF interface structure
The OSPF interface structure is described in Section 9 of
[OSPF]. In a MOSPF router, the following items are added to the
OSPF interface structure. Note that the IPMulticastForwarding
parameter is really a description of the attached network. As
such, it should be configured identically on all routers
attached to a common network; otherwise incorrect routing of
multicast datagrams may result[13].
o IPMulticastForwarding. This configurable parameter indicates
whether IP multicasts should be forwarded over the attached
network, and if so, how the forwarding should be done. The
parameter can assume one of three possible values: disabled,
data-link multicast and data-link unicast. When set to
disabled, IP multicast datagrams will not be forwarded out
the interface. When set to data-link multicast, IP multicast
datagrams will be forwarded as data-link multicasts. When
set to data-link unicast, IP multicast datagrams will be
forwarded as data-link unicasts. The default value for this
parameter is data-link multicast. The other two settings are
for use in the special circumstances described in Sections
6.3 and 6.4. When set to disabled or to data-link unicast,
IGMP group membership is not monitored on the attached
network.
o IGMPPollingInterval. When the router is actively monitoring
group membership on the attached network, it periodically
sends IGMP Host Membership Queries. IGMPPollingInterval is a
configurable parameter indicating the number of seconds
between IGMP Host Membership Queries. The router actively
monitors group membership on the attached network when both
a) the interface's IPMulticastForwarding parameter is set to
data-link multicast and b) the router has been elected
Designated Router on the attached network. See Section 9 for
details.
o IGMPTimeout. This configurable parameter indicates the
length of time (in seconds) that a local group database
entry associated with this interface will persist without
another matching IGMP Host Membership Report being received.
See Section 9 for details.
o IGMP polling timer. The firing of this interval timer causes
an IGMP Host Membership Query to be sent out the interface.
The length of this timer is the configurable parameter
IGMPPollingInterval. See Section 9 for details.
8.3. Additions to the OSPF neighbor structure
The OSPF neighbor structure is defined in Section 10 of [OSPF].
In a MOSPF router, the following items are added to the OSPF
neighbor structure:
o Neighbor Options. This field was already defined in the OSPF
specification. However, in MOSPF there is a new option which
indicates the neighbor's multicast capability. This new
option is learned in the Database Exchange process through
reception of the neighbor's Database Description packets,
and determines whether group-membership-LSAs are flooded to
the neighbor. See the items concerning flooding in Section
14 for a more detailed explanation.
8.4. The local group database
The local group database has already been introduced in Section
2.3.1. The current section attempts a more precise definition.
The local group database tracks the group membership of the
router's directly attached networks. Database entries are
created and maintained by the IGMP protocol. Database entries
can cause group-membership-LSAs to be originated, which in turn
enable the pruning of datagram shortest-path trees. The local
group database also dictates the router's responsibility for the
delivery of multicast datagrams to directly attached group
members.
Each entry in the local group database has three components: the
multicast group, the attached network and the entry's age. A
database entry is indexed by the first two components: multicast
group and attached network. A database lookup function is
assumed to exist, so that given a [multicast group, attached
network] pair, the matching database entry (if any) can be
discovered. A database entry for [Group A, Network N1] exists if
and only if there are Group A members currently located on
Network N1.
The three components of a local group database entry are defined
as follows:
o MulticastGroup. The multicast group whose members are being
tracked by this entry. Each multicast group is represented
as a class D IP address. For the semantics of multicast
group membership, see [RFC1112].
o AttachedNetwork. Each database entry is concerned with the
group members belonging to a single attached network. To get
a complete picture of the local group membership (when for
example building a group-membership-LSA), it may be
necessary to consult multiple database entries, one for each
attached network. Note that a router is only required to
maintain entries for those attached networks on which the
router has been elected Designated Router or Backup
Designated Router (see Section 9).
o Age. Indicates the number of seconds since an IGMP Host
Membership Report for multicast Group A has been seen on
Network N1. If the age field hits Network N1's configured
IGMPTimeout value, the local group database entry is removed
(i.e., the entry has "aged out"). See Sections 9.2 and 9.3
for more information.
8.5. The forwarding cache
The forwarding cache has already been defined in Section 2.3.
The current section attempts a more precise definition. Each
entry in the forwarding cache indicates how a multicast datagram
having a particular [source network, destination multicast
group, IP TOS] will be forwarded. A forwarding cache entry is
built on demand from the local group database and the datagram's
shortest-path tree. For more details, consult Sections 2.3.4 and
12.
Each entry in the forwarding cache has six components: the
multicast datagram's source network, the destination multicast
group, the IP TOS, the upstream node, the list of downstream
interfaces and (possibly) a list of downstream neighbors. A
forwarding cache entry is indexed by source network, destination
multicast group and IP TOS. A lookup function is assumed to
exist, so that given a multicast datagram with a particular [IP
source, destination multicast group, IP TOS], a matching cache
entry (if any) can be found.
The six components of a forwarding cache entry are defined as
follows:
o Source network. The datagram's source network is described
by a network/subnet/supernet number and its corresponding
mask. The source network for a datagram is discovered via a
routing table/database lookup of the datagram's IP source
address, as described in Section 11.2.
o Destination multicast group. The destination group to which
matching datagrams are being forwarded. For the semantics of
multicast group membership, see [RFC1112].
o IP TOS. The IP Type of Service specified by matching
datagrams. Note that this means that the path of the
multicast datagram depends on its TOS classification.
o Upstream node. The attached network/neighboring router from
which the datagram must be received. If received from a
different attached network/neighboring router, the matching
datagram is dropped instead of forwarded. This prevents
unwanted replication of multicast datagrams. It is possible
that the upstream node is unspecified (i.e., set to NULL).
In this case, matching datagrams will always be dropped, no
matter where they are received from. It is also possible
that the upstream node is specified as the placeholder
EXTERNAL. This means that the datagram must be received on a
non-MOSPF interface in order to be forwarded.
o List of downstream interfaces. These are the router
interfaces that the matching datagram should be forwarded
out of (assuming that the datagram was received from
upstream node). Each interface is also listed with a TTL
value. The TTL value is the minimum number of hops necessary
to reach the closest (in terms of router hops) group member.
This allows the router to drop datagrams that have no chance
of reaching a destination group member.
o List of downstream neighbors. When the datagram is to be
forwarded out a non-broadcast multi-access network, or if
the interface's IPMulticastForwarding parameter is set to
data-link unicast, the datagram must be forwarded separately
to each downstream neighbor (see Sections 2.3.3 and 6.4). As
done for downstream interfaces, each downstream neighbor is
specified together with the smallest TTL that will actually
reach a group member.
9. Interaction with the IGMP protocol
MOSPF uses the IGMP protocol (see [RFC1112]) to monitor multicast
group membership. In short, the Designated Router on a network
periodically sends IGMP Host Membership Queries (see Section 9.1),
which in turn elicit IGMP Host Membership Reports from the network's
multicast group members. These Host Membership Reports are then
recorded in the Designated Router's and Backup Designated Router's
local group databases (see Section 9.2).
9.1. Sending IGMP Host Membership Queries
Only the network's Designated Router sends Host Membership
Queries. This minimizes the amount of group membership
information on the network, both in terms of queries and
responses.
When a MOSPF router becomes Designated Router on a network, it
checks to see that the network's IPMulticastForwarding parameter
is set to data-link multicast (see Section B.2). If so, it
starts the interface's IGMP polling timer. Then, whenever the
timer fires (every IGMPPollingInterval seconds), the MOSPF
router sends a Host Membership Query out the interface. The
destination of the query is the IP address 224.0.0.1. For the
format of the query, see [RFC1112]. If/when the MOSPF router
ceases to be the network's Designated Router, the IGMP polling