Page 332

306 Chapter 10: Managing Network Performance with Queuing and Compression

NOTE

Notice that the word "queuing" is misspelled as "queueing." The Cisco IOS has contained that
particular typographical error for a number of years.

Example 10-6 notes the default queue and then line-by-line details about the remainder of the
queue configuration. Note that the changes that were made to the queue sizes are given and that
the sizes for the medium and normal queues are not listed because they've been left at the
default settings.

Custom Queuing

Custom queuing enables the sharing of available bandwidth evenly (or not) across all types of
traffic. This technique allocates a percentage of bandwidth to each of the various traffic types.
The difference between this approach and priority queuing is that the queues are processed in
round-robin sequence (in essence, they are multiplexed). Therefore, it is possible that high-
priority traffic would not be serviced quickly enough because although each type of traffic
would get some bandwidth, no traffic would be designated with a higher priority than the rest.

Custom queuing employs 17 queues. Queue 0, the system queue, is used (as the name implies)
for the system. Queue 0 is the only queue that has a higher priority than the remaining queues
and that might not have traffic assigned to it. The traffic serviced by queue 0 is as follows:

ISO IGRP hellos

ESIS hellos

IS-IS hellos

DECnet hellos

SLARP address resolution

EIGRP hellos

OSPF hellos

Router syslog messages

Spanning Tree keepalives

The administrator configures the remaining queues, 116, which are served on a round-robin
basis.

With priority queuing, the lower priority traffic types can starve out and be dropped due to
latency because if the high and medium queues are active at all times, no traffic from the low
and normal queues are serviced. Custom queuing alleviates this condition by allocating
bandwidth evenly.