I'm new to this environment and when I arrived I notices that five catalyst switches are trunk to one 6513 core switch, which is fine. The 6513 switch is the server farm switch and also the aggregation switch. The issue is that sw22 is the root bridge, not the 6513 switch. Sw22 is a user switch that is trunk into the 6513 switch. The issue is that when a user needs to access a server off of the 6513 switch they must first connect into sw22 because sw22 is the root bridge, is that correct? This tells me that the 6513 switch is acting like a transit switch. If this is true then this could be related to the slow response issue at my site. I think the resolution for this problem would be to make the 6513 switch the root bridge, do you agree?
It is not necessarily true that any user must first connect into sw22 because sw22 is the root bridge. It would depend a bit on the topology of the network and would depend on whether the users are in the same VLAN as the server.
Having said that, it would probably be beneficial to make the 6513 be the root of the spanning tree since the 6513 seems to be the center of the network and it is usually beneficial to have the root at the center of the network.
Give it a try and let us know if performance improves.
Some additional information, the 6513 switch VTP is setup as server and the other switches including sw22 the rootbridge are setup as clients. All of the VLANs that exist on the 6513 also exist on the user switches, does that mean that when accessing a server in vlan 200 on the 6513 clients do not have to traverse the rootbridge? I thought that all traffic traverses the rootbridge. My network topology is that all switches including sw22 are trunk into the 6513 switch.
VTP setup with 6513 as server and other switches as clients is fine.
When you talk about a server in VLAN 200 on the 6513 are the users also in VLAN 200? From the fact that the switches are connected by trunks I am going to assume that users are in different VLAN(s) from the server VLAN. In this case you must have inter VLAN routing. Any traffic from a user in some other VLAN must get to a layer 3 interface and be routed to be able to access servers in VLAN 200 (and the response must be routed back). That is likely to have more impact on perfromance than is the question of who is the root of the spanning tree.
I have 16 different VLANs configure and that four out of my five switches are the rootbridge for a few of the VLANs. If I want to make the 6513 the rootbridge for all vlan can I use the MACRO that's used in the 5000 switch? The command is set spantree root 1 (1-being the vlan). Will this command work with the 6513 switch. If this command works I need to do it sixteen times on the 6513 switch.
I agree with Rick that it is preferrable to make the 6513 STP root. This however will only affect the network response in case of link/device failure and is not in itself an explanation for the performance problems as you describe them.
I would rather check for any duplex mismatches and cabling problems to adress the perf-issues.
Checking some interface counters for error rates would be a good starting point.