
Configuring Frame Relay Services
2-20
117376-C Rev. 00
Traffic Control
Frame relay is unreliable. It does not include error recovery mechanisms, relying
instead on the upper layer protocols to detect and retransmit lost frames.
Hub and spoke topologies often present a “big pipe/little pipe” situation. The
central or hub site has a faster link to the frame relay service because it is a point
of concentration for many remote sites: it has a “big pipe,” with, for example, a
T1/E1 connection that can transmit data at rates of 1.54 Mb/s. The remote sites
usually support a much lower line speed, 56 Kb/s to 64 Kb/s, or “little pipe.”
Figure 2-8
illustrates this concept.
Figure 2-8. Big Pipe/Little Pipe Topology
The central site router sends traffic at its available bandwidth, in this example at
T1 rates (1.536 Mb/s in the United States and Canada, 2.048Mb/s elsewhere).
Some switches recognize that the remote site is configured at a lower speed, and
begin to drop frames above the capacity of the remote site router. At the remote
site, the frame relay interface discards frames beyond its available buffer. The
assumption is that the user application detects the lost frames and either
retransmits them or uses a flow control mechanism in the protocol, such as
windowing, to throttle, which means queue, the traffic. But not all applications
have a robust mechanism to deal with lost or out-of-sequence frames.
You can use Bay Networks WAN Compression Protocol (WCP), protocol
prioritization, congestion control, and traffic shaping either singly or in
combination to help control the flow of traffic and avoid loss of data.
FR0013A
64 Kb/s
64 Kb/s
B
D
64 Kb/s C
1.536/2.048Mb/s
A
Comentarios a estos manuales