Priority Queue

h.zelenka · ‎02-21-2012

I have a Cisco Catalyst 2960 with IOS Release12.2(53)SE (because of a contract I can not update it) -> the release notes for this version describe the following:

When auto-QoS is enabled on the switch, priority queuing is not enabled. Instead, the switch uses shaped round robin (SRR) as the queuing mechanism. The auto-QoS feature is designed on each platform based on the feature set and hardware limitations, and the queuing mechanism supported on each platform might be different. There is no workaround. (CSCee22591)

My config is as follows:

interface FastEthernet0/1
 switchport access vlan 200
 switchport mode access
 srr-queue bandwidth share 10 10 60 20
 priority-queue out 
 mls qos trust dscp
 auto qos voip trust 
 no cdp enable
 network-policy 1
 spanning-tree portfast

My question now is:
When the priority queue is not enabled with auto-qos because of the software bug is it nevertheless 
enabled with the additional priority-queue out command?

Thank You

Peter Paluch · ‎02-21-2012

Hello,

To my best knowledge, the Auto-QoS is just a system of macros that enables specific QoS features by their individual commands so you don't have to, but there is nothing in the Auto-QoS you couldn't do with plain commands. Therefore, even if the Auto-QoS does not add the priority-queue out command automatically, adding it manually should activate the priority queueing on the egress interface without any setbacks.

You may want to verify the state of the interface using the show mls qos int fa0/1 queue command.

Best regards,

Peter

h.zelenka · ‎02-22-2012

Thank you Peter for your reply.
My next question now is:

AutoQos has configured:

 srr-queue bandwidth share 10 10 60 20

The Bug-Report (CSCee22591) describes the following:

When auto-QoS is enabled on the switch, priority queuing is not enabled. Instead, the switch uses shaped round robin (SRR) as the queuing mechanism.

So normally when I configure

srr-queue bandwidth share 10 10 60 20

srr-queue bandwidth shape 10 0 0 0

This config means that Queue 1 has now shaping enabled and queue1 is removed

from the 'sharing' and any packet entering Queue1 will be de-queued first but

shaped at 1/10 of the interface bandwidth.

But what does the description in the Bug-Report mean?
Does it mean one queue uses shaped round robin instead of a priority queue
or does it mean all queues use shaped round robin?

Peter Paluch · ‎02-23-2012

Hello,

I would need to see the entire interface configuration to tell what kind of round robin is used - whether shaped or shared. However, if you enter these two commands:

srr-queue bandwidth share 10 10 60 20

srr-queue bandwidth shape 10 0 0 0

then the queueing behaves as follows:

Queue 1 is guaranteed and shaped (i.e. it is not ever allowed to exceed) to the 1/10 of the egress port throughput
Queue 2 has a guaranteed ratio of 10/(10+60+20) = 10/90 = 1/9 of the egress port throughput
Queue 3 has a guaranteed ratio of 60/(10+60+20) = 6/9 of the egress port throughput
Queue 4 has a guaranteed ratio of 20/(10+60+20) = 2/9 of the egress port throughput

Queues 2-4 can also exceed their guaranteed rate if there is a free capacity.

If the total allocation exceeds 100% of the egress port capacity, the shaped queue is given preferential treatment, consuming the total remaining capacity for the remaining shared queues.

But what does the description in the Bug-Report mean?
Does it mean one queue uses shaped round robin instead of a priority queue 
or does it mean all queues use shaped round robin?

It probably means that the command priority-queue out is not added automatically to an interface, in which case the queues operate in combined shaped/shared way. Please note that there is a difference between shaped/shared queue, and a priority queue. Only the queue 1 can be declared as priority queue, and in that case, it is always served until absolutely empty. Only when empty, the scheduler serves the remaining queues. There is no shaping/sharing performed on a priority queue and it is allowed to consume 100% of the egress interface capacity at all times.

Best regards,

Peter

h.zelenka · ‎02-24-2012

Here ist the output of the sh mls qos int fa 0/10 queue:

cat1#show mls qos int fa0/10 queue
FastEthernet0/10
Egress Priority Queue : enabled
Shaped queue weights (absolute) : 25 0 0 0
Shared queue weights : 10 10 60 20
The port bandwidth limit : 100 (Operational Bandwidth:100.0)
The port is mapped to qset : 1

So Q1 is the Priority Q and has a shaped queue weight of 25 --> 4 Mbps

Queues 2-4 use shared roun robin with weights of 10, 60 and 20 (Peter has explained the exact ratio of the queues already)

Right?

Peter Paluch · ‎02-24-2012

Hello,

Thank you for the output.

So Q1 is the Priority Q and has a shaped queue weight of 25 --> 4 Mbps

Q1 is indeed the prirority queue here, but precisely because of that, both its shaped and shared weights are totally ignored and are irrelevant currently. Q1 is allowed to consume all available bandwidth. If it was not configured as a priority queue then it would be both guaranteed and shaped to 1/25th of the available bandwidth, i.e. 4 Mbps. However, as a priority queue, it has unlimited priority.

Best regards,

Peter

Pronoy Dasgupta · ‎02-23-2012

Hello,

Well if i were to look at the bug report:

http://tools.cisco.com/Support/BugToolKit/search/getBugDetails.do?method=fetchBugDetails&bugId=CSCee22591

It seems that this is going to be a steady feature, and that priority queueing will not be enaled if you enable Auto-QoS.

This is a platform based decision made by the respective business unit and would be not be changed as such.

The other point of whether the queues use shaping or sharing has been explained by Peter already.

If you consider only this line:

srr-queue bandwidth shape  10  0  0  0

It simply implies that queue 1 is to be shaped only, the rest of the other queues would be in sharing mode.

HTH

Pronoy

mendotecsa · ‎03-31-2013

Hello,

We have Avaya IP Phones 9611G on Catalyst 2960S Gigabit PoE, and use the auto qos voip trust.

I wish to know two hing about auto-qos:

1) why the Auto-QoS configure on the global config all parameters about queue-set 1:

mls qos queue-set output 1 threshold 1 100 100 50 200

mls qos queue-set output 1 threshold 2 125 125 100 400

mls qos queue-set output 1 threshold 3 100 100 100 400

mls qos queue-set output 1 threshold 4 60 150 50 200

mls qos queue-set output 1 buffers 15 25 40 20

and then on the interface port the auto-qos autoconfigures this:

queue-set 2

2) What exactly the switchport voive vlan does? Because if only the VoIP traffic is correctly marked, we are sure that the VoIP traffic will queued in the quie 1 (expedit)...

thanks

mendotecsa · ‎03-31-2013

Hello,

We have Avaya IP Phones 9611G on Catalyst 2960S Gigabit PoE, and use the auto qos voip trust.

I wish to know two hing about auto-qos:

1) why the Auto-QoS configure on the global config all parameters about queue-set 1:

mls qos queue-set output 1 threshold 1 100 100 50 200

mls qos queue-set output 1 threshold 2 125 125 100 400

mls qos queue-set output 1 threshold 3 100 100 100 400

mls qos queue-set output 1 threshold 4 60 150 50 200

mls qos queue-set output 1 buffers 15 25 40 20

and then on the interface port the auto-qos autoconfigures this:

queue-set 2

2) What exactly the switchport voive vlan does? Because if only the VoIP traffic is correctly marked, we are sure that the VoIP traffic will queued in the quie 1 (expedit)...

thanks