Difference in LLQ implementation between GSR/IOS-XR and other platforms

[narainarun]

Hi All,

I have a problem in understanding how LLQ is implemented in different platforms of Cisco.

Idealy QoS should kick in only when there is a congestion in the link irrespective of queueing / scheduling (LLQ and CBWFQ).

But in certain platforms like GSR and IOS-XR, LLQ is confiugred only with priority and police command not with "prioirity percent <value>" command. In priority and police command since policer is used, LLQ is always on even there is no period of congestion. Of course with police you can re-mark the exceed traffic to different marking but thats not the requirement in my case.

In platforms like 7206, LLQ is configured with "prioirty percent <value>" which works ideally only when there is a period of congestion. When there is no congestion, LLQ class can use scanvenge other classes as well.

Would like to know is there any specific reason why there is a difference in the implentation of LLQ between different platforms of Cisco.

thanks in advance

Arun

[narainarun]

Hi All,

I found the difference in LLQ implementation among different platforms.

There are two types of implementation with respect to LLQ.

Congestion aware LLQ - This is configured with "priority " command. This allows LLQ use other classes when there is no congestion in the link. This is supported in software based platforms like 7200.

Always ON LLQ - This is configured with "priority and police" command. This allows LLQ not use other classes even there is  no congestino in the link. This is supported in hardware based platforms like GSR, CRS, 7600. Though Cisco says this is the recommended method, hardware based platforms only support "Always ON LLQ".

thanks

Arun