[c-nsp] Fabric Enabled Vs. Non-Fabric Enabled.

[Drew Weaver]

Hello all,

I have a couple of 6513s in my lab with WS-X6348-RJ-45, this is not a fabric enabled card. I have read that the older non-fabric enabled cards suffer from issues with input output buffer overflow which leads to queue drops and packet loss, etc.

My question is, how do you know if the fabric enabled/non fabric enabled (and thus the 128k packet buffer memory on the WS-X6348-RJ-45) is causing the IQD and OQD counters to go up on show int sum and not another problem like a counter bug?

And are there ever any other types of things that you've seen which can cause that counter to increment besides packet loss? Such as BPDUFILTER, etc?

I know this discussion has been beaten to death but I can't really find any actual concrete evidence that says that the reason for those counters incrementing is because of the packet buffer size on the cards.

I have no problem upgrading the cards, but I need to make sure that the issues I am seeing when pushing test traffic will be diminished by upgrading.

I am running Version 12.2(18)SXD7b and the show platform capacity command is apparently not included in this release.

Thanks,
-Drew