spanning-tree gateway load balancing (STP & GLBP)

Hello, packethief! You wrote on 12 Mar 2007 18:46:53 -0700:

p> Can someone tell me if having the same VLAN that spans multiple p> switches and connects to dual distributions really works with p> GLBP??? I have two interconnected distribution switches p> (Distribution-A and Distribution-B) that have multiple access p> layer switches (access-A & access-B) connected to each p> distribution switch. Each access layer switch has the same VLAN p> (Vlan2) across both switches. . Please take a look at this PDF: p>

p> On page 40, figure 41 - with VLAN 2 present on two access layer p> switches, and both switches are uplinked to the two distribution p> switches. p> In this configuration, wouldn't one of the uplinks on Distribution p> switch B be blocked not only on the link that connects to p> distribution- A and also to one of the access layer switches. p> I don't see how GLBP can work if you have dual distribution p> switches connected to multiple access layer switches that span p> Vlans. This document seems wrong

Good catch.

p> Can someone correct me if I?m wrong.

No, you are correct. On a bright side, this picture makes somewhat nasty inteview question :-)

With best regards, Andrey.

PDF:

On page 40, figure 41 - with VLAN 2 present on two access layer

I do not see a problem with the diagram. In this case, blocking between the cores prevents two loops....one in the Core 1 / Core 2 / Access 1 network, and one in the Core 1 / Core 2 / Access 2 network. By making it block between the cores (only on the VLANs that you need GLBP), your arps will be heard by both cores, and you enable your two cores to load balance with 2 virtual macs rather than 1 (as it shows on page 38). Why would a core switch need to block on another link out to the access layer? You don't have connections between the access layer switches, so I don't see another loop that needs a block?

What am I missing?

Hello, Trendkill! You wrote on 13 Mar 2007 04:54:35 -0700:

T> I do not see a problem with the diagram. In this case, blocking T> between the cores prevents two loops....one in the Core 1 / Core 2 T> / Access 1 network, and one in the Core 1 / Core 2 / Access 2 T> network. By making it block between the cores (only on the VLANs T> that you need GLBP), your arps will be heard by both cores, and T> you enable your two cores to load balance with 2 virtual macs T> rather than 1 (as it shows on page 38). Why would a core switch T> need to block on another link out to the access layer? You don't T> have connections between the access layer switches, so I don't see T> another loop that needs a block?

T> What am I missing?

Loop - Core1->Access1->Core2->Access2->Core1.

With best regards, Andrey.