Network Performance

[benson.low]

Hi CSC,

I am intrigued by this area. We have a few catalyst switches, mainly 2950 series, with Version 12.1(13)EA1 IOS. Using jperf, we are seeing drops in graph. See attached jpg.How to solve this issue?

Next issue is that people are complaining of slow file access on a file server, i have checked the switchport and no errors were registered, except for

underruns and out buffer failures. What does this mean?

  5 minute output rate 185000 bits/sec, 162 packets/sec
     996217351 packets input, 3104837961 bytes, 0 no buffer
     Received 544266 broadcasts, 0 runts, 0 giants, 0 throttles
     0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
     0 watchdog, 0 multicast, 0 pause input
     0 input packets with dribble condition detected
     1527706738 packets output, 2547903229 bytes, 1062040 underruns
     0 output errors, 0 collisions, 1 interface resets
     0 babbles, 0 late collision, 0 deferred
     0 lost carrier, 0 no carrier, 0 PAUSE output
     1062040 output buffer failures, 0 output buffers swapped out

[Leo Laohoo]

Do a wireshark trace.  Most of the time if the switch can't see anything then there's something wrong on the server end.

Look at the server event logs if you can find something funny.

[Latchum Naidu]

Hi Roger,

I would suggest you check the 2950 switches CPU & is these have more uptime? If yest try to reboot them once.

Regarding file server slow access....Check the cable between switch to fileserver I would suggest cat6 cable to be used. If possible change the cable with new one. And check the resource utilization of the file server which is very important.

Please rate the helpfull posts.
Regards,
Naidu.

[benson.low]

Hi Naidu,

We have switches running for 7 years without any reboot and we do not face such an issue. If rebooting is the way to go, i guess Cisco is very much align to "commercial" OS way of operations, , reboot and it will solve the problem. Do not mistaken me, I am a ardent fan of Cisco and I believe Cisco produces quality products and services. What I am trying to point out is that, such issues can only be identitfied using throughput tools. However, we do not go around testing every switches, just to ascertain that it is still working fine. There must be a more effective and systematic way of approaching this issue. Any way of doing this?

[benson.low]

Hi,

Thanks for the response. We are the LAN management team, and server management is beyond our administrative. What is frustrating, we are NOT given enough information on server side. I guess wireshark is the next best way to go.

[Leo Laohoo]

We are the LAN management team, and server management is beyond our administrative.

I share your pain.  I've been in your exact situation.

In 9 out of 10 cases, wireshark will find the culprit.  In the majority of our issues, wireshark will tell us that the network will receive the packets from the clients to the server but the server takes forever to return the query.