Inter-VLAN Problem on 6509 MSFC? (Slow response from one VLAN)

[jmarimo]

I have two Catalyst 6509 switches one of which recently failed and I am therefore using the remaining one for all my connections. I have a number of VLANs defined and am routing between them via the MSFC.

My problem is this: when I do a file transfer from one of the VLANs to a remote machine across our WAN link (a metro Ethernet 100Mbps link) the response time appears slow from a specific VLAN but appears OK on another. When I do basic or extended PING tests, latency appears fine, but when I do bandwidth testing via FTP, it appears there is a bandwidth issue

For example, from the production VLAN, (which has about 80 machines connected) to transfer a 50Mb file via FTP across the 100Mbps metro link takes 14 seconds but to do a similar transfer from the test and staging VLAN (which has only 3 machines connected), it takes about 7 seconds.

Any ideas on what may be causing this problem? So far, I have had a look at host/platform configuration issues (the source machines are a combination of Windows 2000 and Solaris servers, and the destination machine is a Solaris Informix database server) but cannot come up with anything that may be causing the issue.

I have spoken to our Metro Ethernet link provided and they have checked the link and confirmed that there are no problems with the link, it is configured for 100Mbps, full duplex, at either end

I can send you our config if that helps ... Many thanks for your help in anticipation.

Jonathan

[netops01]

Jonathan

I have seen a similar problem on my 6506/MSFC-2 setup. After much head scratching, I deleted the VLAN ant layer 3 and 2. Rebooted switch and recreated VLAN using same IP etc. The problem dissapeared. I realise there is a service outage but this was acceptable to my users as I did it out of core hours.

Ian

[jmarimo]

Ian,

Thank you very much for your prompt response, unfortunately this option is a rather drastic one as the switch is in production. I will have to see if I can do it on my other (as yet) non-production switch. Will let you know how it goes. Anyone else come across the problem and used a different approach?

Many thanks,

Jonathan

[jmarimo]

Ian,

Thank you very much for your prompt response. Unfortunately ours is a production switch which makes it difficult to find downtime to re-build the switch from scratch as a solution.

Anyone else experienced the same problem but used a different approach to solve it? I will try and see if I can re-build my other non-production switch and see if I can `cure' the problem that way.

Many thanks,

Jonathan