This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: CNT5106C Computer Networks, Summer 2010 Instructor: Prof. Ahmed Helmy Homework #5 On Multicast at the Network Layer (IP Multicast and Membership) Q1. (10 points) In several multicast routing protocols we use RPF check (reversed- path forwarding check). What is the purpose of such check? How does it work? What are the underlying assumptions this check relies on? A1. The purpose of RPF check is to avoid loops in multicast routing. We want to build a tree, which is loop-free, with the sender serves as the root. Hence a router only accepts incoming multicast traffic when it comes from the interface that is used to send packet toward the senders IP address on unicast routing table. By doing so a router only accepts incoming multicast traffic from each sender on one interface, hence avoids the possibility of forming routing loops. The underlying assumptions of RPF check are: (1) It depends on unicast routing table, so unicast routing table must be correct and converged for RPF checks to work properly. (2) It assumes that the path used from a sender to a router and the reversed path from the router back to the sender are symmetric. If they are not, RPF check would reject multicast traffic on the shortest path from the sender to the router. It leads to non-optimal multicast tree. [Note: in cases where the links are uni-directional, then the reverse path approach can fail altogether.] Q2. (10 points) What are the differences between the targeted environments (potential number of group members etc.) for PIM-DM and PIM-SM? How does this lead to different protocol design? A2. In PIM-DM it assumes that most parts of the network want the multicast traffic. Therefore it starts with a flooding to build a complete multicast tree and later removes unwanted branches. Since most parts of the network want the multicast traffic, it makes sense to have a routing state on all the routers for each group, as it would be useful with high probability. In PIM-SM, on the contrary, it assumes that most parts of the network do not want the multicast traffic. In stead of initial flooding, specific join mechanism is used in PIM- SM, and states are only created on routers that must know the state to forward the multicast traffic. Q3. (18 points) IGMP provides membership information to the first hop router regarding the existence of receivers on a directly connected LAN. (a) Why are group-specific query messages introduced in IGMPv2? Argue showing what a router does when it receives a leave message from a host. (b) The multicast host model does not define any interaction between the sources and IGMP. Do you see any problems with that? Explain. [Hint: Think of a case where there are no members for a group.] (c) A researcher suggested to have the source multicast a query to the group in order to find out whether there are members in the group or not. Receivers would respond to this query by sending response to the source....
View Full Document
This document was uploaded on 05/27/2011.
- Spring '09