connecting two 6000 switches

[sreid]

I have a catalyst 6009 with two Supervisor engines and I will soon be getting another 6000 with a supervisor engine probably two. My question is can anyone give me advice on how to connect the two. Can I get a couple of GBICs and use the ports on the supervisor modules or should 10/100 ports setup with etherchannel? I guess basically my question is, do the ports on the supervisor engines act the same as ports on other modules?
Has anyone done anything similar? Any ideas are much appreciated.

thx

 

[KiscoKid]

Yeh I've used the Gigabit ports that come with the Supervisor module before. There is absolutely no reason why you cannot configure this as either a trunk or access link and run Etherchannel over it.

 

[webnetwiz]

If these are your Core switches, then you don't want to run trunks between them, just connect them with 2 glass links (make sure STP is running).

 

[Piass]

webnetwiz - why wouldn't you setup trunks between your core switches?

sreid - I would go with KiscoKid's advice.

As this will become an important link, it would be advisable to use 2 connections and etherchannel them together. If you do this, use one port on each supervisor. Then, if one of the supervisor's fails, you will not lose you connection.

This is a very common scenario.

 

[webnetwiz]

Ok, time for a design lesson. Cisco's states the following - "Cisco's philosophy discourages trunking across the core, therefore discrete VLAN1 connections are made between the distribution and core switches". Let's step back and see. There is no Layer 3 activity that needs to go on in the core. The core is strictly Layer 2, for the purpose of fast switching. If you carry trunks between your core switches, then you carry all the unnecessary VLANs and other traffic, something that in a large environment has no purpose to travel through the core.
Now, I'm not saying trunks won't work, but why carry everything else, if all you need is VLAN1? All I'm saying is that if you're designing a Campus Network, that's the recommended design from Cisco. Doesn't mean anyone has to follow their reference design, but since they manufacture and design their equipment, have the necessary labs to actually build and test all of these technologies in practice, shouldn't we at least look and see if they have a point? Maybe their model is the one that is most suitable if you're trying to design a robust and resilient campus network?
All things to consider. I was suggesting, looking at sreid's post, that it looks like his network would be a candidate for the switchblock design.

Piass, maybe you want to consider reading through CiscoPress's Building Cisco MultiLayer Switched Networks, ISBN# 1578700930.