- Information
- AI Chat
Was this document helpful?
Distance Vector Multicast Routing (DVMR)
Course: Systems Programming (01:198:214)
23 Documents
Students shared 23 documents in this course
University: Rutgers University
Was this document helpful?
Distance Vector Multicast Routing (DVMR)
Each router is aware, given a unicast routing table, that the current shortest route to a certain
destination passes through NextHop. Thus, whenever the router receives a multicast packet from
source S, it only transmits the packet if and only if it arrived through the link that has the shortest
path to S on all outgoing links (other than the one on which it arrived) (i.e., the packet came from
the NextHop associated with S in the routing table). While not looping packets back toward S,
this tactic effectively floods packets out from S. This method has two significant drawbacks. The
first is that there is no mechanism for avoiding LANs where there are no other members of the
multicast group, therefore it really floods the network. Below, we address this issue.
The second restriction is that any router linked to a LAN will forward a specific packet over that
LAN. This is because, regardless of whether those links are a member of the shortest-path tree
rooted at the source, the forwarding technique of flooding packets on all links other than the one
on which the packet arrived. Eliminating the redundant broadcast packets that are created when
multiple routers are linked to a single LAN will address the second issue. The parent router is the
only router that is permitted to forward multicast packets from that source over the LAN, so one
approach to achieve this is to designate one router as the parent router for each link relative to the
source. The parent router is chosen based on which router has the shortest path to source S; if
there is a tie between two routers, the router with the smallest address wins.
Based on the distance-vector messages it exchanges with its neighbors, a particular router can
determine if it is the parent for the LAN (again in relation to each potential source). It's important
to note that this improvement mandates that each router maintain a bit for each of its incident
links for each source to indicate whether or not it is the parent for that source/link pair. As an
internet router is solely concerned with forwarding packets across networks, keep in mind that in
an online environment, a source is a network rather than a host. Reverse Path Broadcast (RPB)
and Reverse Path Forwarding are two names for the resulting technique (RPF). In contrast to
unicast routing, which seeks out the shortest way to a specific destination, we are evaluating the
shortest path toward the source while making our forwarding selections, so the path is reversed.
Shortest-path broadcast is implemented by the RPB method that was just discussed.
We now wish to remove the networks that do not have any hosts that are G members from the list
of networks that receive each packet directed to group G. Two steps can be taken to complete
this. We must first be able to tell when a leaf network is devoid of group members. If the parent
router mentioned above is the only router on the network, then the network is a leaf, which can
be easily determined. As explained in our earlier discussion of link-state multicast, finding out if
any group members are present on the network is performed by periodically having each host
that belongs to group G announce this information over the network. Using this knowledge, the
router can then select whether or not to forward a multicast packet to G via this LAN.
The "no members of G here" information is propagated up the shortest-path tree in the second
step. To do this, the router adds the set of groups for which the leaf network is interested in
receiving multicast packets to the hDestination, Costi pairs it delivers to its neighbors. Once this
knowledge has spread from router to router, each router will know which groups to pass
