QoS Bandwidth Allocation

[hotpackets]

I'm looking at a customer's QoS configuration, and I noticed that they identified bandwidth, instead of percentage, which adds up to something much less than the interface bandwidth. I assume that the remaining bandwidth is shared in proportion to the configured rates of each queue. 

Let's say I have the following queues:

• Q1 - Priority 1Mb
• Q2 - Bandwidth 10Mb
• Q3 - Bandwith 29Mb
• Default - Bandwidth 5Mb

Total 45Mb on a 100Mb link.

So the actual allocation would be this:

• Q1 - 2.22% - 2.22Mb
• Q2 - 22.22% - 22.22Mb
• Q3 - 64.44% - 64.44 Mb
• Default - 11.11% - 11.11 Mb

Assuming I'm on the right track:

• Q1 would have a guaranteed maximum of 2.22Mb.
• Q2 would have a guaranteed minimum of 22.22Mb.
• Q3 would have a guaranteed minimum of 64.44Mb.
• Default would have a guaranteed minimum of 11.11Mb.

Do I understand this correctly? 

[paul driver]

Hello

Are you relating this to quality of service on switches?
Can you elaborate a little on what devices this is relating to and possible post configuration to assist in your query?

[hotpackets]

Hi Paul. It's an ASR1K. Config would look something like this:

policy-map QUEUE
 class EF
  priority 1000
 class AF31
  bandwidth 10000
 class AF11
  bandwidth 29000
 class class-default
  bandwidth 5000

interface FastEthernet1/1
 service-policy output QUEUE

Not really looking for a validation of configs here. Just need to know what happens when queues are allocated a total of 45m on a 100m interface.