28 Chapter 1: Quality of Service: An Overview
• A class called Customer-D will get AF41, AF42, and AF43 traffic.
For each customer, an ingress policer marks the traffic based on the rate of the traffic. Up
to 256 kbps is marked AFx1, above 256 kbps but below 768 kbps is marked AFx2, and
above 768 kbps is marked AFx3. During periods of congestion, this traffic is placed into the
classes described earlier as it egresses the aggregation router toward other regions. Because
each class of traffic is independently forwarded, and each class is guaranteed a certain
amount of bandwidth (using CBWFQ), the classes do not interfere with each other in any way.
Assume congestion and that Customers A, B, and C are all sending traffic at their CIR, but
Customer D is sending traffic at 1.5 Mbps. It wouldn’t be fair to punish the other customers
because Customer D is not “behaving,” so traffic is queued in the class called Customer-D
and, when that queue begins to fill, WRED begins to discard packets. Per the RFC that
defines AF, packets marked AF43 (those that were over 768 kbps) are discarded first, then
packets marked AF42 (those that were between 256 kbps and 768 kbps), and packets
marked AF41 are discarded only if no packets are marked AF43 or AF42.
It’s not difficult to see how this could provide a service similar to Frame Relay where there
is a CIR with burst and extended burst capabilities. However, no hard rule mandates the
purpose for which AF must be used in your network.
Summary
This chapter defined QoS as the ability to create predictable service levels for various traffic
types in the network, and also as “managed unfairness” (that is, the ability to provide different
traffic types with unequal treatment while giving each type the treatment that it requires).
Integrated services and differentiated services have been discussed in detail, and you should
now understand where you might find each of these useful in your network. The thorough
discussion of some of the key DiffServ RFCs has provided the background information that
you need in future chapters, which discuss Cisco’s specific implementations of these
standards. As these mechanisms are discussed, you should be able to relate their behavior
to the standardized behaviors explained by the various RFCs that have been discussed in
this chapter.
This page intentionally left blank This page intentionally left blank
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.