CITRIX QOS

[islam.irshaid]

Dear all,

Could anyone please help me with CITRIX(Meta frame V.4) QOS, Knowing any PDLM Version supported.

The QOS that I applied show that citrix was not detected from either the head office and branches,but it detect eigrp, rtp voice control, mgcp.

and please tell me if "ip nbar protocol-discovery" command is Necessary or not in this case ?

Please find attached the Configuration file.

[royalblues]

Are you encrypting the Citirx traffic using IPSEc?

In this case you need to use qos-preclassify command under the crypto map to copy the TOS contents onto the Ip header

http://www.cisco.com/warp/public/105/crypto_qos.html

HTH

Narayan

[islam.irshaid]

Thank you , but i am using the classification and marking on the remote site and no need to pass Through VPN tunling as TOS.

the problem is no match on remote site, i think is a PDLM problem, since i matched the HTTP,Skiny , ..

Regards

Islam

[mheusing]

Hi,

"ip nbar protocol-discovery" was needed in older IOS versions, but this requirement has been removed. I forgot the exact IOS version though, when this was done, but as you indicate that match protocol gives you matches for other protocols, you should be fine.

Can you please check the ports CITRIX is using in your environment and compare them to "show ip nbar port-map"? Citrix traffic can only be detected by NBAR, if the port matches the (pre-)configured one.

In case you are using different ports, then you can adjust that with configuring "ip nbar port-map citrix ...".

Hope this helps! Please rate all posts.

Regards, Martin

[jwdoherty]

Narayan's recommendation is excellent, some other points, though.

I see the marker is on a parent FastEthernet rather than the subinterfaces. Check that you are seeing Citrix, and other hits, within the marker service policy on that interface. Besides Martin's suggestion to confirm the port usage, I wonder whether the policy should be applied to the subinterfaces.

I don't see any bandwidth allocation for class 25?

I see you're using FQ in the class-default. Just be aware on non-7500 platforms, every FQ flow competes for bandwidth against the other defined classes. This might not be what you intended.

I also see you're using WRED in the class-default. The default's early drop policy default settings are usually not advantageous for non-TCP traffic.

There is a PDLM for Citrix that can match against the Citrix's internal priority. Don't recall whether the later 12.4 images have that version. Useful if you want to treat Citrix "screen scraping" traffic different from Citrix printing or file copying.

Lastly, you might consider whether the MTU should be adjusted to account for IPSec overhead. (Often helps improve throughput performance.)