Congestion control

9

consume the full buffer memory. Thresholds can also be used to bound the maximum queueing delay for each traffic class. Additionally if the sum of the thresholds for a port is set below 100 percent of the buffer memory, then you can also ensure that a single port does not monopolize the entire shared memory pool.

FIGURE 9

Queue depth

The tail drop algorithm can be extended to support per priority drop thresholds. When the ingress port CoS queue depth breaches a threshold, then any frame arriving with the associated priority value will be dropped. Figure 9 describes how you can utilize this feature to ensure lower priority traffic cannot totally consume the full buffer memory. Thresholds can also be used to bound the maximum queueing delay for each traffic class. Additionally if the sum of the thresholds for a port is set below 100 percent of the buffer memory then you can also ensure that a single CoS does not monopolize the entire shared memory pool allocated to the port.

Changing the Tail Drop threshold

Perform the following steps from Privileged EXEC mode to change the Tail Drop threshold. 1. Enter global configuration mode.

switch#configure terminal

2.Change the Tail Drop threshold for each multicast traffic class. In this example, 1000pkt is used.

switch(config)#qos rcv-queue multicast threshold 1000 1000 1000 1000

Example of increasing multicast frame expansion Tail Drop Threshold to 1000pkt for each multicast Traffic Class.

switch:admin>cmsh switch>enable

switch#configure terminal

Enter configuration commands, one per line. End with CNTL/Z.

switch(config)#qos rcv-queue multicast threshold 1000 1000 1000 1000 switch(config)#end

3. Enter the copy command to save the running-config file to the startup-config file.

Converged Enhanced Ethernet Administrator’s Guide

99

53-1001761-01

 

Page 121
Image 121
Brocade Communications Systems 53-1001761-01 manual Changing the Tail Drop threshold