Collapsed Backbone

[PAUL CHEN]

Hi, I have a few questions pertaining to a collapsed backbone design.

L3switch1 ----- L3switch2

Two L2 switches with redundant uplinks to each L3 switch. The uplinks will be 802.1q trunks.

Question : If only static routes are allowed in the L3 switches, would it be possible to have a 802.1q inter switch trunk between the two L3 switches ?

If this is done, would spanning tree lead to the blocking of this trunk or the L2 switch uplinks.

I know this sounds like a bad design but would this ever work ?

Any comments appreciated.

Thanks

[ericgarnel]

The trunks would perform independently of the L3 functionality.

Yes, STP will block one of the trunks, and this is a good thing. It prevents bridging loops.

It also depends on the link type & cost.

If the link between the L3 switches is a gig and the trunk links from the l2 to l3 switches are fast ethernet, then STP will chose the lowest cost.

It is not necessarily a bad design, that really depends on your environment.

There is a more to STP than what can be said here. Here is a good starting point:

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_configuration_example09186a008009467c.shtml

here is an example:

South: primary root for odd vlans & subnets

North: secondary root for odd vlans & subnets

reverse the roles for even numbered vlans & subnets

So for vlan 9, the south L2 switch will be root and the north l2 switch will be root secondary. for subnet 192.168.9, the south l3 switch will be hsrp active, and the north l3 switch will be hsrp standby.

and for vlan 10, the north L2 switch will be root and the south l2 switch will be root secondary. for subnet 192.168.10, the north l3 switch will be hsrp active, and the south l3 switch will be hsrp standby.

[royalblues]

It is actually a good design.

Configure etherchannel between the L3 switches and force any one switch as the root switch and the other as backup root switch.

Since all the access switches are redundantly connected to the core, STP will block one of the uplinks to the core to avoid loops.

You can also configure HSRP or MHSRP for gateway redundancy. These HSRP hellos will be sent over the L2 connection between the L3 switches

HTH, rate if it does

Narayan