Network redesign questions

Unanswered Question
Mar 12th, 2009


We are going to be replacing our mixed Dell and Cisco switching environment with an all Cisco solution as you can see in the attachment. We consolidated a great many of Dell and Cisco switches in our data center with the 4510R-E we recently purchased.

What we have now are two Dell switches with all access layer switches directly connected to both core switches via fiber. Using VRRP, one does all interVLAN routing using the directly connected VLANs. If one core dies, the other router takes over via VRRP.

We cannot do the Cisco routed layer 3 design so what I plan to do is pretty much the same thing but use HSRP for the redundancy component. I have two questions for our new design...

1) Is there any advantage or reason that we would want to implement a routing protocol into this network?

2) If we go the HSRP route, is there any reason that the two core switches should not be directly connected?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
glen.grant Thu, 03/12/2009 - 09:19

If that the whole network you shouldn't need a routing protocol , if you have other sites its connected to then probably.

No reason why you can't connect the 2 switches , set your spanning tree roots to match the hsrp active side and it should put the second link out of the access switch into a blocked state . Its a pretty common design and we have a big site pretty much like that and they are trunked between the two main L3 devices. Been working fine for years .

roitsupport Thu, 03/12/2009 - 09:50

Good deal, this now brings up my next question...

Currently we implement MST for redundant links and it works well. With the new setup I would like to implement etherchannel between the two core switches. What is the best way to set this up and how does STP interact with link aggregation?

Giuseppe Larosa Thu, 03/12/2009 - 11:19


>> how does STP interact with link aggregation?

all STP types treat an etherchannel as a single logical link allowing the usage of all member links.

So no problem on using an etherchannel between the two core switches it improves performance and fault tolerance.

There are some safety measures that have to be used:

before setting up the bundle the configuration of all member links is to be identical:

speed, duplex, native vlan, trunk encapsulation, list of permitted vlans

after the bundle is up and working if you need to do any change like adding a vlan in the list of permitted vlans do it on the port-channel logical interface : the necessary changes are then replicated on member links automatically.

Trying to change the list of permitted vlans of a member link of a working bundle can create serious problems like bridging loops

Hope to help



This Discussion