Campus Network Design Help

Hello,

I am previewing a network run at a hospital and I could use a suggestion or two on its design.

Keep in mind that since it is a hospital, it is imperative that it operate in full redundant mode, which it MOSTLY does.

There are 16 closets maintaining one or two access layer switches. Each of these switches are connected via redundant VLAN trunks to two distribution switches (6509).

The 6509s are L3 switching to the core as well as to their medical partner. The issue is at this time there is only one connection to the partner through the first switch. At one point they maintained a trunk connection between the first and second distros, but now maintain a L3 GigEtherChannel. EIGRP is using four different vlans to route traffic to the partner. If data is coming in to the second switch, the data flow is then being passed to one or more access layer switches and then back up to the first distro switch and finally out to the partner.

The problem is that the hospital maintains at least 100 different VLANs throughout the access/distro layers. Some of these vlans have access-lists associated with them. If EIGRP decides to choose one of these VLANs to forward data, time-sensitive applications break.

So, the questions:

1) How should the distro switches really be connected, VLAN Trunk or L3? 2) Should passive-interface be installed on every single VLAN to prevent choosing paths through closet switches? 3) What about upping the bandwidth between the distros to 10 Gig to choose that link over the trunk links?

Thanks

Robert

Reply to
myhrer
Loading thread data ...

Hello, snipped-for-privacy@stjosephs-marshfield.org! You wrote on 9 Feb 2005 10:18:34 -0800:

m> So, the questions:

m> 1) How should the distro switches really be connected, VLAN Trunk m> or L3?

It depends. Do you span more than one access switch with any given VLAN? If yes, than trunk is your only option. If no than you can use V topology. Clark's Cisco LAN Switching book has an excelent capter on campus design.

m> 2) Should passive-interface be installed on every single VLAN to m> prevent choosing paths through closet switches?

Yes, yes and yes. You don't want to have peering through any available VLAN. Depends on topology one or two would be more than enough.

m> 3) What about upping the bandwidth between the distros to 10 Gig m> to choose that link over the trunk links?

Without network diagram it's very hard to tell what would and wouldn't help. So far I have an impression that you don't have a clean design out there. It might help to think about Layer 2 and Layer 3 connectivity separately. I normally treat any Layer 3 switch as essentially two boxes - normal layer 2 switch and router on a stick. SVI interfaces would be carried on a link between router and L2 box, routed interface would belong to router itself. You can further decouple it to VLAN level.

With best regards, Andrey.

Reply to
Andrey Tarasov

Thank Andey,

I needed to go to lunch because I no more got out the door and it occured to me that they have to have a trunk because they are spanning VLANs across multiple switches. Because of this, the only VLANs that should need to be passive-interface are those that have actual access-lists on them.

I suppose I could passive interface all but the specific VLANs that should be used to do any routing to the partner site.

Reply to
myhrer

Drop it egghead, and she should nb the word lie in the definition.

They say opposites attract. I hope you meet someone who is good-looking, intelligent, and cultured.

-- Lady Chatterly

"Hi Lady C, I think you fit in well here on Usenet. I recently asked someone in our group if he was a multiple identity of another poster and he started swearing at me. I guess he thought that made him more real. See you around." -- stevejdufour

Reply to
Lady Chatterly

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.