Re: 6500 poor LAN performance

mike-t · ‎04-25-2002

I recently upgraded my LAN to a Catalyst 6500 with gigabit fiber to 3500xl switches. Now whenever anyone copies a large file 50MB+ to or from a server, the LAN "crashes". Some or all workstations have packet loss, or complete disconnects from apps, servers, etc.

The LAN is divided into VLANs....1 for desktops, 1 for servers, and 1 for routers, firewalls, etc. Each VLAN experiences the same issues when a workstation in either VLAN copies the file. (either in the same VLAN or to another VLAN)

Has anyone ever ran into this situation with a similar setup?

DELL ACORD · ‎04-25-2002

I take it your trunking to the 3548's. If so what encapsulation are you using? ISL or dot1Q? Do you have STP enabled and if so are your trunks in the forward state?

DTA

mike-t · ‎04-25-2002

yes i am trunking to the 3500's

ISL encapsulation

STP is enabled

yes the trunks are in the forward state

DELL ACORD · ‎04-25-2002

Are there any errors in the sys log like duplex mismatch or excessive collisions where the servers are connected too? I di have an issue where my 3COM NIC cards caused some issues on my servers where they had to be hardcoded at both the switch and server end to 100/full to eleviate my slowness as well. Another thing that I had to do is that there is an issue with 3548's and some switches that 802.1Q must be used. I had to do this between a 4006 and a 3548XL to fix some packet drops too!!!

DTA

mike-t · ‎04-25-2002

no errors, infact, the NIC's on the server and connected ports were set to 100/full at the start of the project.

The entire LAN is made up of Cisco Equipment (routers, switches, etc) so there are no mismatch problems. The only time we experience any issues is when a file is copied. other then that, no errors, bandwidth problems etc. a real head scratcher.

I'll check into the switches though (we're using 3524XLs and a 6509). Maybe i'll have to change to 802.1Q

david.d · ‎04-25-2002

Can you stress your network during some off peak period using software or simply rcp'ng or copying large files? Also, does this come up when copying server to server on the same VLAN or does it always appear to be from one VLAN to another?

You may have mentioned but are you running native or hybrid on your 6509? You running Sup1a or Sup2?

Can you describe what happens during the file transfer? Does the file copy complete successfully? Is it more like when utilization goes high on a flat network so all hosts experience a sever slowdown?

I am running each of my IDFs (stacks of 3 or 4 3548s) on their own subnet/VLAN with a home run gig fibre to the 6509 and the network seems pretty clean. I’m running a fairly old but stable CatOS on the switch and 12.0.7 on the MSFC.

Keep posting, I’m interested in this issue.

rfroom · ‎04-25-2002

Be absolutely sure that the switch ports are not reporting any errors on ports connected to the servers, workstations and other switches. You may want to look at possible NIC driver upgrades on the servers and workstations. I would also check out the release notes for the 6500 and 3500xl to verify that you are not running into a known issue. Taking several sniffer traces on diffrent VLANs and ports may help in diagnosing the problem.

Here the URL that documents known NIC to Switch Issues:

http://www.cisco.com/warp/customer/473/46.html

datnguyen · ‎04-30-2002

Are there two MSFCs in the 6500? If so, make sure each VLAN interface on the MSFC has unique IP address, and use HSRP address for your default gateway. I have seen this problem before with dual MSFC.