[c-nsp] CoPP and WRR

[Antonio Soares]

Thanks for the feedback. It seems it's a bit more difficult than I thought.
For example, by default, the 6704-10GE uses a combination of Tail Drop and
WRED:

Router#sh queueing interface tenGigabitEthernet 2/1
<...>
    queue tail-drop-thresholds
    --------------------------
    1     70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    2     70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    3     100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    4     100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    5     100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    6     100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
    7     100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 

    queue random-detect-min-thresholds
    ----------------------------------
      1    40[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8] 
      2    40[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8] 
      3    70[1] 70[2] 70[3] 70[4] 70[5] 70[6] 70[7] 70[8] 
      4    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      5    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      6    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      7    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 

    queue random-detect-max-thresholds
    ----------------------------------
      1    70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      2    70[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      3    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      4    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      5    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      6    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 
      7    100[1] 100[2] 100[3] 100[4] 100[5] 100[6] 100[7] 100[8] 

    WRED disabled queues:      4  5  6  7
<...>

So in order to make queue 1 behave like the default (non qos) fifo queue, I
think we need something like this:

no wrr-queue random-detect 1
wrr-queue threshold 1 100 1 1 1 1 1 1 1 

The final configuration for each 10GE interface would be:

Router#sh run int te2/1 
Building configuration...

Current configuration : 263 bytes
!
interface TenGigabitEthernet2/1
 no ip address
 shutdown
 wrr-queue bandwidth percent 100 0 0 0 0 0 0 
 wrr-queue queue-limit 100 0 0 0 0 0 0 
 wrr-queue threshold 1 100 1 1 1 1 1 1 1 
 no wrr-queue random-detect 1 
 wrr-queue cos-map 1 1 0 1 2 3 4 5 6 7 
end

Router#


Does it make sense ?

Thanks.

Regards,

Antonio Soares, CCIE #18473 (R&S/SP)
amsoares at netcabo.pt
http://www.ccie18473.net


-----Original Message-----
From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Phil
Mayers
Sent: terça-feira, 3 de Setembro de 2013 13:39
To: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] CoPP and WRR

On 03/09/13 13:23, Saku Ytti wrote:
> On (2013-09-03 13:10 +0100), Antonio Soares wrote:
>
>> wrr-queue bandwidth percent 100 0 0 0 0 0 0
>>   wrr-queue queue-limit 100 0 0 0 0 0 0
>>   wrr-queue cos-map 1 1 0 1 2 3 4 5 6 7
>>
>> But doing this kind of stuff for hundreds of interfaces doesn't make 
>> too much sense.
>
> Alas it is what you must do. The problem is that default QoS isn't 
> this, it should be. But Cisco tries to be helpful in Catalyst BU and 
> offer some magic default QoS which I'm certain causes more issues than it
solves.

Yeah, it's a shame you can't set the global defaults - it's a lot of typing,
and slows down the already-too-slow NVGEN on these platforms.
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/