[c-nsp] Question about CBWFQ and PING times

[Andy Saykao]

Hi Peter,

Thanks for the detailed reply. I forgot to include the router platforms
we are using for this.

[GROUP1] --> [ POP1] <--> [POP2] --> [HOSTED SERVICES + INTERNET]

POP1 = Cisco 7204VXR (NPE-G1) GigE Interface running 12.2(31)SB13

POP2 = Cisco 7606 with 4-subslot SPA Interface (7600-SIP-400) running
12.2(33)SRB3

1/ "If you have a 200mbps connection going out from GigabitEthernet-link
your prioritising won't take effect, since buffers will never saturate."
So if we were to prioritize something like Voice, we would need to
implement some Heirarchical QoS solution on the Cisco 7606???

2/ I understand your arguments exactly about not prioritizing ICMP
traffic, but I was told to look into this. I guess based on 1/ above,
some form of Heirarchical QoS solution is needed for this also.

Cheers.

Andy


-----Original Message-----
From: Peter Rathlev [mailto:peter at rathlev.dk] 
Sent: Wednesday, 25 March 2009 12:38 PM
To: Andy Saykao
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Question about CBWFQ and PING times

Hi Andy,

On Wed, 2009-03-25 at 11:15 +1100, Andy Saykao wrote: 
> 1/ We have a 200mb link between two POPS that is being congested in 
> the evening. Congestion is happening on the outbound direction from 
> POP2 to POP1, so from a user's perspective in GROUP1 it would be 
> impacting their download.
...
> interface GigabitEthernet4/0/2
>  service-policy output POP2-POP1-QOS-POLICY
> 
> I can see matches for this when doing a show policy-map interface. Is 
> it as simple as this to ensure that users in GROUP1 will be assured of

> bandwidth to access our hosted services?

If you have a 200mbps connection going out from GigabitEthernet-link
your prioritising won't take effect, since buffers will never saturate.
Heirarchical QoS (as discussed thoroughly many times recently on this
list) with a parent shaper could solve this, but it is uncertain if your
platform can do this. What hardware and IOS version are you using?

Another possibility would be to police some of the traffic that causes
the congestion, which even the least feature rich switches with L3
features can do.

If you have some SRR-device (Catalyst 3560, 3750, some 6500 modules) you
could do some crude shaping, but the number of queues available often
makes this an interesting task and traffic limited this way could be an
unpleasant experience for the users.

> 2/ If I wanted to prioritze ping times between POP1 to POP2, how would

> this be done?

On a side note: Giving priority to ICMP Echo is in my eyes a bad
strategy. This is almost by definition not important business traffic,
so the main reason to give it higher priority would be to avoid
problematic questions from incompetent users who only know how to
measure latency and loss this way. It will not give them a better
experience in using the network connection itself and might instead hide
certain symptoms that could be helpful in troubleshooting one day.

Regards,
Peter



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

This email and any files transmitted with it are confidential and intended
 solely for the use of the individual or entity to whom they are addressed. 
Please notify the sender immediately by email if you have received this 
email by mistake and delete this email from your system. Please note that
 any views or opinions presented in this email are solely those of the
 author and do not necessarily represent those of the organisation. 
Finally, the recipient should check this email and any attachments for 
the presence of viruses. The organisation accepts no liability for any 
damage caused by any virus transmitted by this email.