Quality of Service: An Overview
The constantly changing needs of networks have created a demand for sensitive applica-
tions (such as voice over IP (VoIP) and video conferencing over IP), and networks are being
asked to support increasingly mission-critical data traffic. Providing predictable service
levels for all of these different types of traffic has become an important task for network
administrators. Being able to provide predictable and differentiated service levels is key to
ensuring that all application traffic receives the treatment that it requires to function properly.
This chapter covers several aspects of quality of service (QoS) and discusses how to provide
QoS in Cisco networks. Specifically, this chapter covers the following topics:
Understanding QoS
Deploying QoS in the WAN/LAN: High-Level Overview
Cisco AVVID (Architecture for Voice, Video, and Integrated Data)
Overview of Integrated Services and Differentiated Services
Differentiated Services: A Standards Approach
Understanding QoS
The following section defines QoS in terms of measurable characteristics. It is important,
however, to recognize that fully understanding QoS requires more than a definition. To truly
understand QoS, you must understand the concept of managed unfairness, the necessity for
predictability, and the goals of QoS. In addition to a definition of QoS in measurable terms,
the following section explains each of these things, to provide you with a well-rounded and
practical definition of QoS.
Definition of QoS
QoS is defined in several ways, and the combination of all of these definitions is really the
best definition of all. A technical definition is that QoS is a set of techniques to manage
bandwidth, delay, jitter, and packets loss for flows in a network. The purpose of every QoS
mechanism is to influence at least one of these four characteristics and, in some cases, all
four of these
6 Chapter 1: Quality of Service: An Overview
Bandwidth itself is defined as the rated throughput capacity of a given network medium or
protocol. In the case of QoS, bandwidth more specifically means the allocation of
bandwidth, because QoS does not have the capability to influence the actual capacity of any
given link. That is to say that no QoS mechanism actually creates additional bandwidth,
rather QoS mechanisms enable the administrator to more efficiently utilize the existing
bandwidth. Bandwidth is sometimes also referred to as throughput.
Delay has several possible meanings, but when discussing QoS, processing delay is the
time between when a device receives a frame and when that frame is forwarded out of the
destination port, serialization delay is the time that it take to actually transmit a packet or
frame, and end-to-end delay is the total delay that a packet experiences from source to destination.
Jitter is the difference between interpacket arrival and departure—that is, the variation in
delay from one packet to another.
Packet Loss
Packet loss is just losing packets along the forwarding path. Packet loss results from many
causes, such as buffer congestion, line errors, or even QoS mechanisms that intentionally
drop packets.
Table 1-1 shows examples of the varying requirements of common applications for
bandwidth, delay, jitter, and packet loss.
Managed Unfairness
Another, more pragmatic definition of QoS is managed unfairness. The best way to explain
this definition is using an analogy to airline service. Sometimes airlines are unable to sell
all of their seats in first class, but they rarely leave the gate with first class seats available,
Table 1-1 Traffic Requirements of Common Applications
Voice FTP Telnet
Bandwidth Required Low to moderate Moderate to High Low
Drop Sensitive Low Low Moderate
Delay Sensitive High Low Moderate
Jitter High Low Moderate

Get Cisco Catalyst QoS: Quality of Service in Campus Networks now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.