background image
304 Chapter 9: WAN Design Considerations
To help you understand the role that bandwidth plays, let's take a look at the function of hot
water in the shower:
In an apartment building, all the tenants share much of the same plumbing infrastructure. When
all the tenants request hot water at the same time, the demand for hot water often outweighs the
available supply. With only so much hot water available, someone has to take a cold shower or
wait until hot water becomes available. The same is true of bandwidth.
When all the users request bandwidth at the same time, the demand for bandwidth often outweighs
the available supply. With only so much bandwidth available, someone has to wait until bandwidth
becomes available. Because network applications are continuing to grow and consume more
bandwidth, Cisco has developed tools to ensure that bandwidth is conserved and that mission-critical
data reaches its destination on time. Cisco's QoS tools allow complex networks to control when an
application's data will arrive. Every network that carries mission-critical traffic needs QoS for
optimum efficiency. QoS speeds the process of handling mission-critical applications while sharing
network resources with noncritical applications. QoS also ensures available bandwidth and
minimum delays required by time-sensitive multimedia and voice applications.
Priority Queuing and Custom Queuing
Many enterprises run Cisco networks with a mixture of SNA and client/server protocols.
Networks in the near future will be designed to run voice, video, and data. Networks have
traditionally operated on a best-effort delivery principle, which means that all traffic has equal
priority and an equal chance of being delivered in a timely manner until congestion occurs. As
a result, when congestion occurs, all traffic has an equal chance of being dropped. A key
challenge when designing internetworks is to preserve the predictable response of sensitive
traffic while ensuring that the network maintains its throughput requirements, even during
periods of congestion.
If you think traffic volume or bandwidth limitations could cause performance degradations in the
design of your network, this section is a must-read. Although attaching higher-speed lines to
ensure enough bandwidth is a solution and an obvious remedy, it is not always cost-effective. A
cost-effective design solution would implement queuing. Please keep in mind that the queuing
techniques described in this chapter do not take effect unless there is congestion in the network.
Priority Queuing
Priority queuing gives strict priority to important traffic and guarantees that important traffic gets
the fastest handling at each point where it is used. Priority queuing prioritizes according to
protocol, incoming interface, packet size, and source/destination address. Each packet is placed
in one of four queues based on an assigned priority: high, medium, normal, or low. High-priority
traffic is always preferred over the other queues. When carrying time-sensitive traffic such as voice
or mission-critical traffic such as SNA, priority queuing is preferred. If your design uses voice
traffic, a priority queue would guarantee preferential treatment. When congestion occurs on the
WAN, priority queuing guarantees that voice traffic will be sent first and foremost. When the voice
queue empties, traffic on the next-highest-priority queue is transmitted.
87200333.book Page 304 Wednesday, August 22, 2001 2:53 PM