Design Reference
Table Of Contents
- Contents
- Chapter 1: Introduction
- Chapter 2: New in this release
- Chapter 3: Network design fundamentals
- Chapter 4: Hardware fundamentals and guidelines
- Chapter 5: Optical routing design
- Chapter 6: Platform redundancy
- Chapter 7: Link redundancy
- Chapter 8: Layer 2 loop prevention
- Chapter 9: Spanning tree
- Chapter 10: Layer 3 network design
- Chapter 11: SPBM design guidelines
- Chapter 12: IP multicast network design
- Multicast and VRF-lite
- Multicast and MultiLink Trunking considerations
- Multicast scalability design rules
- IP multicast address range restrictions
- Multicast MAC address mapping considerations
- Dynamic multicast configuration changes
- IGMPv3 backward compatibility
- IGMP Layer 2 Querier
- TTL in IP multicast packets
- Multicast MAC filtering
- Guidelines for multicast access policies
- Multicast for multimedia
- Chapter 13: System and network stability and security
- Chapter 14: QoS design guidelines
- Chapter 15: Layer 1, 2, and 3 design examples
- Chapter 16: Software scaling capabilities
- Chapter 17: Supported standards, RFCs, and MIBs
- Glossary
Multiple user mode allows several users on the same port or VLAN. If one user leaves the
group and other receivers exist for the same stream, the stream continues. The switch tracks
the number of receivers that join a given group. For multiple user mode to operate properly,
do not suppress reports. This ensures that the switch properly tracks the correct number of
receivers on an interface.
The Fast Leave feature is particularly useful in IGMP-based TV distribution where only one
receiver of a TV channel connects to a port. In the event that a viewer changes channels
quickly, you create considerable bandwidth savings if you use Fast Leave.
You can implement Fast Leave on a VLAN and port combination; a port that belongs to two
different VLANs can have Fast Leave enabled on one VLAN (but not on the other). Thus, with
the Fast Leave feature enabled, you can connect several devices on different VLANs to the
same port. This strategy does not impact traffic after one device leaves a group to which
another device subscribes. For example, you can use this feature when two TVs connect to a
port through two set-top boxes, even if you use the single user mode.
To use Fast Leave, you must first enable explicit host tracking. IGMP uses explicit host tracking
to track all source and group members. Explicit host tracking is disabled by default. For
configuration information, see Avaya Virtual Services Platform 4000 Configuration — IP
Multicast Routing Protocols, NN46251–504.
Last member query interval tuning
If an IGMPv2 host leaves a group, it notifies the router by using a leave message. Because of
the IGMPv2 report suppression mechanism, the router is unaware of other hosts that require
the stream. Thus, the router broadcasts a group-specific query message with a maximum
response time equal to the last member query interval (LMQI).
Because this timer affects the latency between the time that the last member leaves and when
the stream actually stops, you must properly tune this parameter. This timer can especially
affect TV delivery or other large-scale, high-bandwidth multimedia applications. For instance,
if you assign a value that is too low, this can lead to a storm of membership reports if a large
number of hosts are subscribed. Similarly, assigning a value that is too high can cause
unwanted high-bandwidth stream propagation across the network if users change channels
rapidly. Leave latency also depends on the robustness value, so a value of two equates to a
leave latency of twice the LMQI.
Determine the proper LMQI value for your particular network through testing. If a very large
number of users connect to a port, assigning a value of three can lead to a storm of report
messages after a group-specific query is sent. Conversely, if streams frequently start and stop
in short intervals, as in a TV delivery network, assigning a value of ten can lead to frequent
congestion in the core network.
Another performance-affecting factor that you need to be aware of is the error rate of the
physical medium. For links that have high packet loss, you can find it necessary to adjust the
robustness variable to a higher value to compensate for the possible loss of IGMP queries and
reports.
In such cases, leave latency is adversely impacted as numerous group-specific queries are
unanswered before the stream is pruned. The number of unanswered queries is equal to the
robustness variable (default two). The assignment of a lower LMQI can counterbalance this
effect. However, if you configure the LMQI too low, it can actually exacerbate the problem by
inducing storms of reports on the network. LMQI values of three and ten, with a robustness
IP multicast network design
116 Network Design Reference for Avaya VSP 4000 February 2014
Comments? infodev@avaya.com