cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
213
Views
0
Helpful
2
Replies

Small Campus Design (which option?)

a.dean
Level 1
Level 1

I am upgrading our switched network. I have a small campus and will have a collapsed core design. I have 2 options as speced our by our Cisco vendor.

Option #1

1, 6500 core switch with redundant supervisor modules. The access layer switches (6500s also) in the closets could have Gigabit Etherchannel links to the core. The Cisco document concerning 6500 High Availability explains that scenario well.

Option #2

2, 6500 core switches with the closet switches having a link to each. This seems kind of expensive and complex for an additional level of redundancy.

1. I assume one of my uplinks from the closets would be in blocking mode to avoid STP loops, thus I lose the GEC load sharing

I have with option #1. Is this right?

2. I'm using an external router for routing VLANs now on 1 core switch. Must I use a MSFC2 card in both switches and use HSRP for failover routing?

3. What can I do with my WAN router? It has only 1 ethernet interface. Is this a manual switch-over in case of one core switch failure? Is there any other option? I'm willing to purchase additional ahrdware here, if necessary.

Does the chassis of a 6509 fail often? What if I just bought a spare 6509 chassis and some cards and kept them around. I'm sure we could withstand 2 hours downtime to swap out a chassis.

It seems like the redundant switches for option #2 offer no load balancing features and a failover would never be totally automatic. Anyone have thoughts on this?

2 Replies 2

efrahim
Level 4
Level 4

both options should work but in case of option 2 you redundancy is not depending on one core switch. Regarding the blocking policy in option 2, if you don't have layer connection between two core switches , then there 'll be no blocking but then each access layer switch should only belongs to vlans that are not common on other access switches.

If you are only one interface on the wan router ,you can do much, other options is to buy another ethernet module if possible, like if you have 3600 routers.

hope this helps

7rbowenii
Level 3
Level 3

1. I assume one of my uplinks from the closets would be in blocking mode to avoid STP loops, thus I lose the GEC load sharing I have with option #1. Is this right?

This depends on your configuration. It is difficult to make recommendations without knowing the number of users/servers you are supporting. If this is a small campus, do you really need 6500 series switches? If cost is important (an it usually is) there may be a better (read more affordable) solution.

If you use two vlans per closet, you can split your users between the two. Set stp primary root for one vlan on one link and the other link primary root on the other vlan. This will get you some load balancing.

You could also connect your closet switch utilizing a Gigabit Etherchannel. This will increase your bandwidth out of your closet, and the bundle within the GEC acts as one physical link. You would not lose this capability if you use spanning-tree.

2. I'm using an external router for routing VLANs now on 1 core switch. Must I use a MSFC2 card in both switches and use HSRP for failover routing?

They key word here is must. You do not have to use a MSFC to route. You can continue to route with the router you have. Now, for fast routing/switching a MSFC is a good solid option. You can configure HSRP between your routing devices for failover. You could also configure HSRP such that it provides some manually configured load-balancing.

3. What can I do with my WAN router? It has only 1 ethernet interface. Is this a manual switch-over in case of one core switch failure? Is there any other option? I'm willing to purchase additional ahrdware here, if necessary.

You could use your WAN router to provide a third redundant HSRP peer. That's really up to you. I try to seperate the WAN from the campus type connections at least via layer-3 (if not layer 2) to limit problems with broadcast storms, runaway processes and the like.

Does the chassis of a 6509 fail often? What if I just bought a spare 6509 chassis and some cards and kept them around. I'm sure we could withstand 2 hours downtime to swap out a chassis.

I have had 6509s on hand for two-ish years, and I can't recall an actual chassis failure. We do keep extra cards on hand in case there is a need to swap one out, but that's it. If this is a large concern of yours, perhaps you should look into purchasing maintenance from Cisco. They have some pretty decent programs with quick turnarounds. I believe the longest I have had to wait for a part is about 2 hours, and our contract says 4.

It seems like the redundant switches for option #2 offer no load balancing features and a failover would never be totally automatic. Anyone have thoughts on this?

You should be able to manually configure several load balancing and redundancy features in either setup.

If you want to discuss in more detail, e-mail me at rbowen@erac.com

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco