Multi-Interface VLAN

Are the two switches isolated for that vlan, or do they connect at some point?

The switches present the same layer-2 VLAN... The idea is to have redundant interfaces into a VLAN. Doing this with etherchannel (F5 == Trunking with LACP) is not a problem. But etherchannel doesn't run ACROSS a switch. Only within it.

 

 

Doing it WITH spanning tree SHOULD be possible. But we'd really rather not run spanning tree (In fact we're running with portfast enabled so there should be no STP running).... What I need to know is what's the defined bahaviour for an F5 when you configure a VLAN with two interfaces (Trunked or not trunked) with regards to bridging traffic and ARP responses (e.g. It's defined that you either have a VMAC or the VLAN will use the lowset numbered interface MAC. But nothing says which interface will be used to respond to an ARP request recieved on both interfaces. Neither does anything say what happens if one interface goes down (I'd expect the F5 to just use the other. But I'd prefer it was documented somewhere. (Maybe it is & I just haven't found it of course)).

 

 

The vlan interface should respond to all arp requests regardless of which interface it is received on if spanning-tree is disabled.

 

 

I doubt this is documented, btw.

That would be fine... What I want to ensure (And my boss wants me to guarantee of course) is that the F5 isn't going to start bridging any traffic...

 

 

H

If spanning-tree is disabled, the LTM will bridge traffic between the two interfaces. I'd definitely mock this up in the lab so you can trend behavior prior to implementation. Let me know if you don't have lab equipment and I'll try to set this up at lunch. You can email me offline if you have specific test criteria at jrahm at charter dot net

I don't have a lab I can do this oin at the moment... Thanks for the offer... What I'd like to see is

 

 

1. What happens if you TRUNK two interfaces without enabling LACP... (Because LACP won't work across switches The switches don't support it)

 

2. What happens if you simply add two interfaces into a VLAN...

 

 

With respects to availability (i..e what happens if one link goes down) and bridging (If a packet comes in on one interface will it be echo'd out the other). Simply bridging the interfaces would be bad...

 

 

H

Would this setup meet both your criteria? I'm still a little fuzzy on your requirements.

Hmm... The trunk/channel between the F5 and the switch really needs to be to two separate switches... (Which is why I can't test it).

 

 

Maybe I need to do some pilfering of switches...

 

 

(The aim is to get redundency of interfaces. Which etherchannel/trunking nornally gives you, but Cisco's are limited to channeling within a switch. And not across switches. Bridging of traffic is not an option...

 

 

H

As long as the switches the F5 connects to are not connected anywhere else, this is fine. If they are, spanning tree is a requirement or you will get bridging loops. You are covered with spanning-tree. This gives you the redundancy you require if one link should fail.

Ah yes. herin lies my dilemma...

 

 

We don't want to run spanning tree from the F5's... We'd rather they just didn't act as a bridge. However although you can disable spanning tree, you can't seem to influence whether the F5 bridges across multiple interfaces in a vlan. And when trunking instead of just using multiple interfaces, there's no docs to say what happens if LACP isn't utilised (ie Does it just fall back to active/standy?)

 

 

There doesn't seem to be any clear docs at this level of detail.

 

 

H