02-07-2007 02:28 PM - edited 03-03-2019 03:40 PM
Hello,
I have applied CBWFQ config to 7200 router ATM pvc, but when I send 2 test files of 200Mbit from the LAN port over the ATM pvc, one from class LBE (lower than best effort using CS1) and one from class default, QoS is not working.
Looking at the stats using 'policy-map int atm 1/0.188 vc 100/192' the packets have been transmitted in class CS1 ok, but it also appears that no packets/bytes have been matched.
Please see my cut down config, showing the 2 classes, policy map and show commands.
Has anybody got any ideas what I have done wrong?
Would appreciate some help. Thanks in advance.
regards
mark
02-07-2007 07:41 PM
I think if you need strict prioritization with your QoS, you should use Low Latency Queueing. With CBWFQ, the QoS will not take effect until the link is congested. It seems that you have a lot of bandwidth and the link is not yet congested that's why it's not in effect.
http://www.cisco.com/en/US/products/sw/iosswrel/ps1830/products_white_paper09186a0080087a84.shtml
"To characterize a class, you assign it bandwidth, weight, and maximum packet limit. The bandwidth assigned to a class is the guaranteed bandwidth delivered to the class during congestion."
HTH
Pat
02-08-2007 12:59 AM
Pat,
The link is a 25Mbps ATM pvc. I did fully congested the link by sending one 200Mbit file from class LBE (CS1) and at the same time one 200Mbit file from default class.
Both arrived (via ftp) in around 18sec. The default class therefore did not receive any more bandwidth than the LBE class.
regards
mark
02-08-2007 05:59 AM
Has anybody else got any idea why the CBWFQ congestion management is not working?
thanks
mark
02-08-2007 06:22 AM
Hello,
looks like you did not overload the interface.
CBWFQ will only care about packets QUEUED, i.e. if no overload condition exits, then no packets are "matched". TCP will adjust to the available bandwidth ... thus the interface might not have been overloaded. Basically then you do not test your CBWFQ config.
On the other hand, 200 MB and 18 seconds give about 90 Mbps!?! So I am not sure about your measurements.
First imho you should check, whether a SINGLE download could potentially fill the pipe, i.e. achieve the full throughput of 25 Mbps.
Additionally UDP flooding the link would make sure there is an overload condition.
Regards, Martin
02-08-2007 07:29 AM
Hi Martin,
I'm keeping you busy at the moment, hope you are well.
The files I sent were 200Mbit not Bytes in size. Thus 2 x 200Mbit files sent together down 25Mbit line arriving in 18sec is reasonable.
I have already tried the same policy on a serial line and tested with suitable size files (via FTP), and it all tested as expected.The TCP windowing did not affect my results.
Thanks again
Mark
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide