blocking vs non-blocking architecture

[jeff.k]

I'm having a hard time finding the distinction between these two on CCO.

I've been told that blocking vs non-blocking is essentially the same as oversubscription. If the backplane doesn't support enough bandwidth for the attached blade/module ports, then those ports are blocking ports. If the backplane does support enough bandwidth for the ports in that blade (or the total number of blades on the switch), then the ports are non-blocking.

Someone else has told me that there are hardware differences beyond bandwidth capacity of the backplane that distinguish blocking from non-blocking.

I'd be grateful for an explanation of these terms and the distinction between them and oversubscription. Links to CCO are also appreciated.

Thanks,

Jeff

[Prashanth Krishnappa]

Blocking vs non-blocking is essentially the same as oversubscription. For example, in a CAT4006 switch, each slot gets 6gig connection to the Supervisor Engine through the backplane. We have a 6 port gig card which is non blocking for all 6 ports while we have 18/24/48 gig port blades which are blocking for most ports