11-02-2011 09:26 AM - edited 03-04-2019 02:08 PM
Hi community,
I've got a problem with our NPE-G2 again. Now I get the packet loss for traversing traffic if I configure a sampled netflow on one of the interfaces.
The router doesn't show too much CPU usage - 40-50%, peaks may be 60%, the overall traffic is about 70-80 Mbit on each of 3 interface - not so much for G2. After I enable the sample netflow on an interface (ir doesn't metter which one), I can see on one interface too many input queue drops and the traffic packets are getting lost. Cisco support means that it could be that we reach the platform limitation, because the enabling of netflow force the router to use process-switching path, and not CEF. But we have many other routers but with G1 which get twice or more traffic and Netflow enable - they have no problems with a input queue drops and packet loss, and their CPU loads are not more then 20%.
Is there any diference in how G1 and G2 process the Netflow data?
11-02-2011 09:48 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Is there any diference in how G1 and G2 process the Netflow data?
Could very well be, as I believe G1 and G2 use different processors and different IOS images.
11-03-2011 03:24 AM
ok, it's correct that G1 and G2 have different CPU and IOS version, but it's stated that G2 has double of performance and the same features as G1, but in real life is not. I can't accept that G2 can process only about 100 Mbit of traffic and G1 has no problems with 300-400 Mbit, it's not fair, G2 costs double
Or may be I'm using somehow wrong the Netflow on G2?
11-09-2011 01:13 AM
just to follow the discussion,
I'm going to update the G2 to a new release train, currently we're using SB, and I'd like to try a "T" train from 12.4.
Is there a possibility that after IOS update the problem with performance in case of Netflow and MPLS somehow will be solved?
11-09-2011 01:40 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Possible solution? I think it could be it the software's implementation is radically "better". Whether its likely is difficult to know. If you're in the position where you can try it, do think it's worth trying.
What IOS version do you use on your G1s that don't exhibit the same problem?
11-09-2011 01:46 AM
Hi Josepf,
it's not a realy possible solution but we have now only 2 options: try new IOS and see or change it to a G1.
G1 has a completly other IOS - they are not compatible with G2.
Our "good old" G1's use the same train "SB" and version 12.2-31-SB18, G2 has 12.2-31-SB13
11-09-2011 02:03 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
I wasn't suggesting trying to run the same physical image. I was curious what IOS version was being used on the G1s. However, since the G1s do have a later patch level, SB18, you might first try that version, if it's available, on the G2. If I recall correctly, the SB version adds features/improvements with their later releases. If there is a patch level later than SB18, you might try that too. I assume there's some reason why SB is being used rather than a mainline release.
Lastly, instead of using 12.4T you might consider instead using 15.0, since that's what 12.4T "morphed" into (I believe).
11-09-2011 02:14 AM
Hi Josef,
thank you for your advices!
SB train was used mainly because of "historical" issues on our PPPoE aggregator. I don't really thing that we need SB train on our edge routers, but I'll try first the latest SB version.
But I think, if were any changes which would change the Netflow and MPLS functions, they would be not implemented inside of the same version/train.
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