Advanced Configuration and Management Guide

Enabling Hardware Forwarding for all Fragments of IP Multicast Packets

By default, an HP routing switch forwards the first fragment of a fragmented IP multicast packet through hardware, but forwards the remaining fragments through the software. You can enable the device to forward all the fragments of fragmented IP multicast packet through hardware.

NOTE: This feature applies only to routing switches, not to switches.

To enable hardware forwarding of all the IP multicast fragments, use the following CLI method.

USING THE CLI

To enable hardware forwarding of all IP multicast fragments, enter the following command at the global CONFIG level of the CLI:

HP9300(config)# ip multicast-perf

Syntax: [no] ip multicast-perf

USING THE WEB MANAGEMENT INTERFACE

You cannot configure this feature using the Web management interface.

PIM Dense Overview

NOTE: This section describes the “dense” mode of PIM, described in RFC 1075. See “PIM Sparse Overview” on page 9-12for information about PIM Sparse.

PIM was introduced to simplify some of the complexity of the routing protocol at the cost of additional overhead tied with a greater replication of forwarded multicast packets. PIM is similar to DVMRP in that PIM builds source­ routed multicast delivery trees and employs reverse path check when forwarding multicast packets.

There are two modes in which PIM operates: Dense and Sparse. The Dense Mode is suitable for densely populated multicast groups, primarily in the LAN environment. The Sparse Mode is suitable for sparsely populated multicast groups with the focus on WAN.

PIM primarily differs from DVMRP by using the IP routing table instead of maintaining its own, thereby being routing protocol independent.

Initiating PIM Multicasts on a Network

Once PIM is enabled on each router, a network user can begin a video conference multicast from the server on R1. When a multicast packet is received on a PIM-capable router interface, the interface checks its IP routing table to determine whether the interface that received the message provides the shortest path back to the source. If the interface does provide the shortest path back to the source, the multicast packet is then forwarded to all neighboring PIM routers. Otherwise, the multicast packet is discarded and a prune message is sent back upstream.

In Figure 9.1, the root node (R1) is forwarding multicast packets for group 229.225.0.1, which it receives from the server, to its downstream nodes, R2, R3, and R4. Router R4 is an intermediate router with R5 and R6 as its downstream routers. Because R5 and R6 have no downstream interfaces, they are leaf nodes. The receivers in this example are those workstations that are resident on routers R2, R3, and R6.

Pruning a Multicast Tree

As multicast packets reach these leaf routers, the routers check their IGMP databases for the group. If the group is not in a router’s IGMP database, the router discards the packet and sends a prune message to the upstream router. The router that discarded the packet also maintains the prune state for the source, group (S,G) pair. The branch is then pruned (removed) from the multicast tree. No further multicast packets for that specific (S,G) pair will be received from that upstream router until the prune state expires. You can configure the PIM Prune Timer (the length of time that a prune state is considered valid).

9 - 4