Overview

When the network gets congested or you start to run out of bandwidth, your QoS policy determines how to allocate the available resources.

In a well-designed network, QoS helps manage every potential bottleneck point. It’s important to implement QoS in the WAN acceleration appliance for the following reasons:

•	It can offload the router.

•	It’s the only element that collects real-time metrics — such as packet loss and delay — for pre-optimization and post-optimization views of the traffic.

In the event that demand exceeds available bandwidth, QoS gives preferential treatment to selected flows, while slowing down or delaying others.

The QoS Policy assigns each flow to a queue that’s associated with a traffic class, for processing and transmission across the WAN:

•	The configuration of the traffic classes determines how likely packets are to get WAN bandwidth at any point in time.

•	Traffic Class definitions are part of the Shaper configuration.

•	The appliance’s WAN interface supports 10 traffic classes.

•	Traffic class definitions, and QoS Policy settings apply to both optimized and pass-through shaped traffic. By default, both share the same limit for maximum bandwidth. However, you can set a lower maximum bandwidth for pass-through traffic than for optimized traffic.

A QoS policy asks:

•	How to you want to use traffic classes to prioritize and shape your traffic?

•	How should the DSCP markings be treated? Trust the incoming LAN or re-mark for the WAN?

•	Do you want to use DSCP markings to prioritize traffic downstream?

The default QoS Policy honors incoming DSCP tags. It also prepopulates the QoS policy table with rules to send traffic to predefined traffic classes (2 - real-time, 3 - interactive, 4 - best-effort) and sends the remaining flows to Traffic Class 1 - default. For the majority of users, the need to adjust this will be a “corner case”.

Please send comments or suggestions regarding user documentation to techpubs@silver-peak.com.