Congestion Avoidance Overview - 3Com MSR 50 Series Configuration Manual

88
Congestion Avoidance
Overview
C
ONGESTION
Excessive congestion can endanger network resources greatly, so some avoidance
measures must be taken. The Congestion Avoidance refers to a traffic control
mechanism that can monitor the occupancy status of network resources (such as
the queues or buffer). As congestion becomes worse, the system actively drops
packets and tries to avoid the network overload through adjusting the network
traffics.
Comparing with the end-to-end traffic control, this traffic control has broader
significance, which affects more loads of application streams through device. Of
course, while dropping packets, the device may cooperate with traffic control
action on the source end, such as TCP traffic control to adjust the network's traffic
to a reasonable load level. A good combination of packet-dropping policy with
traffic control mechanism at the source end can maximize the throughput and
utilization of network and minimize the packet dropping and delay.
Traditional packet-dropping policy
Traditional policy of dropping packets adopts the Tail-Drop method. When the
amount of packets in a queue reaches a certain maximum value, all newly arrived
packets will be dropped.
This kind of dropping policy will lead to phenomenon of TCP synchronization -
when queues drop packets of several TCP connections at the same time, it will
lead these TCP connections to enter congestion avoidance and slow start status to
adjust traffics simultaneously, then reach a high peak of traffics simultaneously. In
this way, network traffic keeps frequent rises and decreases.
RED and WRED
To avoid the phenomenon of TCP synchronization, Random Early Detection (RED)
or Weighted Random Early Detection (WRED) can be used.
In RED algorithm, it sets minimum and maximum limitations for each queue. The
packets are processed as follows:
When the length of queue is less than the minimum limitation, no packet will
■
be dropped.
When the length of queue exceeds the maximum limitation, all the incoming
■
packets will be dropped.
When the length of queue between maximum limitation and minimum
■
limitation, the packet will be dropped randomly. The longer the length of
queue, the higher the dropping probability is, but a maximum dropping
probability will remain.
A
VOIDANCE