Loop Question

[carbonscoring]

Hi, I have an easy loop question for the experienced engineers. Lets say that I had two cisco switches (3550) cabled together twice to each other via Ethernet. The first cable is connected to the identical switch ports vlan. So if I had fa0/1 in vlan 5 to fa 0/1 to vlan 5 in the second switch. And lets say that the same scenario for the second cable just a different vlan. So lets say fa0/2 in vlan 10 to fa0/2 in vlan 10 of the second switch.

Would this cause a loop if I had the typical end hosts on each sides of the switchs using ARP and DHCP?

And if it does cause a loop would STP pick up on this and block one of the switchports?

thanks,

Mike

[ktwaddell]

Hi

Yes it would cause a loop and STP should/would shut 1 of the ports down, doesn't matter it's in a different vlan.

Kev

[Giuseppe Larosa]

Hello Mike,

Cisco STP implementation is per vlan based so both links can be up and running because they are treated as different on the topology of vlan 5 only link1 appears and in the topology of vlan 10 only link2 appears.

This is a legitimate setup unless you start to cross the cables that would cause to join the two L2 vlans/broadcast domains.

Hope to help

Giuseppe

[rducombl]

Actually it depends. in most case it won't cause any loop. it depends of the type of spanning-tree running on the catalyst.

If it in Rapid PVST+ mode or IEEE normal mode, then it won't cause a loop. Indeed in that scenario the switches do build separate L2 topology for each vlan. in your example no port would be blocked, they would all be forwarding . fas 0/1 would allow L2 connectivity for each end host in vlan 5 and fas 0/2 would allow l2 connectivity for host in vlan 10.

In case you would run Spanning-tree MST mode, it may be a bit more complex. If vlan 5 and 10 are part of the same MST instances then indeed spanning would block one port. if they are in separate instances we are back to the same scenario as with Rapid PVST+ mode.

Cheers,

Roland

[carbonscoring]

Thanks for the info everyone!

One other question lets say the same scenario however, the second cisco switch was replaced with a hub/unmanaged switch.

I understand if it was one flat vlan STP would kick in and disable a port.

But, in this situation with separate vlans coming into the hub/unmanaged switch, this would still create a loop in the network and STP would not shut down the ports or see it? If there were DHCP servers in each vlan end host's off the hub/unmanged switch would receive sporadic DHCP IP's from either vlan and frames getting off the hub would work some of the time.

Do I have this correct?

thanks,

Mike

[rducombl]

HI Mike,

Assuming first a hub :

The hub do not run STP but will for sure flood receive BPDU on one port to the other (like it will do for any frame).

So if you connect the hub with 2 ports on the same switch the switch will receive its own bpdu from port 1 to port 2 for example and it will block one of the port (mark as backup port in RSTP terminology). This is true even if both port are in different vlan , so that won't cb.reate a loop still.

Regarding your query with DHCP I am not sure I understand a hub typically do not run vlan so hosts on it will effectively only communicate with device which are located on the switch in the vlan of the forwarding port (making the vlan where the port is blocked useless for user on the hub.

Having said all that connecting with 2 different vlan a switch to a devices which is not Vlan aware or STP aware is probably not a very good idea :)

Roland