QoS is an integral part of any multilayer switched network deployment. With QoS, you can build a network of predictable behavior for latency, jitter, and packet loss. In addition, QoS mitigates anomalous network behavior and provides for differentiation of traffic flows. The following list summarizes QoS:
Classification associates a priority value to a frame.
Catalyst switches support classification on a per-interface or per-packet basis using ACLs.
Marking changes the DSCP value of a packet or CoS value of a frame on ingress or egress.
Catalyst switches use policing or shaping to condition traffic rates.
Catalyst switches support congestion management through the use of WRR, sharing, and shaping scheduling mechanisms.
Congestion avoidance uses WRED on Catalyst switches to improve link utilization under congestion with TCP flows.
In brief, adhere to the following guidelines and recommendations for deploying QoS in a multilayer switched network:
Before configuring Catalyst switches for QoS, develop a QoS plan that clearly outlines the necessary traffic classification.
Opt to classify traffic based on DSCP values instead of CoS values.
Apply classification as close to the edge as possible, preferably in the Building Access submodule.
Trust interfaces that interconnect switches where classification configuration already exists.
Use policing to effectively condition traffic data rates.
Apply congestion management to all interfaces and use priority queuing on interfaces that transmit VoIP traffic.
Use congestion avoidance on interfaces where heavy congestion exists with TCP flows.
For additional information on QoS, refer to the following resources:
"Cisco IOS Quality of Service," at http://www.cisco.com/warp/public/732/Tech/qos/
"Implementing QoS Solutions for H.323 Video Conferencing Over IP," at http://www.cisco.com/warp/public/105/video-qos.html
Cisco Catalyst QoS: Quality of Service in Campus Networks, by Mike Flanagan, Richard Froom, and Kevin Turek (2003; ISBN: 1-58705-120-6)