Thanks for reply. Could you please explain me why the link B is necessary? When some link is fail, e.g. between access1 and distribute1, the packet can traverse from access1 to distribute2 and to core1 switch.

If I misunderstood any points, please advice too.

Thanks a lot,

Nitass

The distribution layer implemtents some kind of policy, control between core and access layer.

If the distribution is not interlinked the distribution layer is partitioned. All traffic that is local to the distribution layer needs to be switched through the core and be processed twice.

Yes the situations outlined in the diagram allows traffic to reach the destionation in case of a failure, however it takes an suboptimal path. So for the sake of performance it's more likely to create a direct connection between the distribution switches.

Leon

Thanks for reply. Do these links depend on the layer which switches will operate? I mean the boundary of layer 2 and 3 of switches.

Thanks a lot,

Nitass

Not nescessarily, since the distribution and core layer consists out of multilayer switches this design is appropriate for multiple configurations.

Also the redundant link between core and distribution is commonly a bigger 'pipe' than the links between access and distribution.

It is common to use 10gig or ethernchannels between the distribution layer switches and the core.

Leon

Thanks for reply. What you exactly mean the link A and B should be there regardless layer which switches will operate. Is it correct?

And I’ve one more question. Could you please advice me the different between layer 2 from core-to-access and layer 3 from core-to-distribution or core-to-access? How can I choose?

Thanks a lot,

Nitass

What I meant that a design that uses links A and B are often found in distribution and core layers.

The access layer is mostly dualhomed between two distribution switches. The distribution layer runs a partial or full mesh with all participating switches in the core.

Where layer two is used in access-layer and terminates in distribution layer 3 is used between core and distribution for high-speed routing.

This limits spanning-tree and VLANs to the appropriate building blocks of Cisco's SAFE implementation.

Thanks for reply. I will think about that.

If I’ve additional question, I will post it again. :)

Have a nice day,

Nitass