Avaya Configuring Frame Relay Services Manual de usuario descargar pdf (Pagina 53)

Frame Relay Overview

308624-14.10 Rev 01

1-31

You can use traffic shaping at one or both ends of a link, but you must use

compression at both ends of a link. It makes sense to use traffic shaping at a

central site, where you have a T1 line that sends data to remote sites with

64 KB line rates, and the goal is to control the flow of traffic and avoid

flooding the remote sites. However, if the CIR is equal to the line rate, which

could well be the case at the remote site with 64 KB line rates, there is no need

to use traffic shaping.

• Compression and traffic shaping together maximize reliability, but at the

expense of throughput.

Traffic shaping occurs at the VC level; compression, at the driver level.

Shaping therefore occurs before compression, which compromises effective

compression because only the precompressed traffic is shaped: the

compressed traffic is not shaped. WCP compresses data at the rate it receives

it, which with traffic shaping is the CIR. It is therefore unrealistic to expect a

high compression throughput for data that originates from a site that also uses

traffic shaping.

Oversubscribing the Traffic Shaping Interface

The CIRs that you configure are based on an average peak rate for the VCs on the

network. If all VCs with traffic shaping try to send data simultaneously, they may

exceed the capacity of the interface. If you oversubscribe the interface, traffic

shaping will still enforce the CIR, but there may be additional latency for reserved

flows. VCs without traffic shaping will send data after traffic-shaped VCs.

Queue Limits and Data Clipping

Nortel Networks routers maintain buffers for each traffic shaped VC. Each buffer

can hold one frame that the router cannot send because of congestion. The router

divides the number of buffers on the interface. Some facts specific to buffers are:

• The default number of buffers is 200 per interface. To change the buffer

number, you can use the Technician Interface.

• The software divides the number of buffers on the interface by the number of

traffic-shaped VCs, giving each VC the same number of buffers. Therefore, if

you have ten traffic-shaped VCs per interface, each VC has 20 buffers.

• If you enable protocol prioritization, the default number of buffers is 30 for

high priority traffic, 200 for normal, and 30 for low.

1 2 ... 48 49 50 51 52 53 54 55 56 57 58 ... 245 246

Sin comentarios

Avaya Configuring Frame Relay Services Manual de usuario Pagina 53