Cisco Support Community
Community Member

Question on ATM interface vs. per vc queueing

I have a question regarding queueing on an ATM interface. Platforms are 7206VXR & 7606s, with either PA-A3 -T3 or PA-A3-OC3SM cards.

What is the functional difference between queueing on the whole ATM interface and per-vc queueing? What would the queue size be for a DS-3 or an OC-3 without per-vc, and with per-vc? What types of symptoms would we see on our PVCs without per-vc queueing?

Also, our "legacy" way of configuring ATM pvcs was to use the old command format:

atm pvc 55 3 55 aal5snap 36864 36864 100

Some on our team have been configuring them with the "new way" as follows:

pvc 35 2/35

vbr-nrt 64 64 1

encapsulation aal5snap

Other than the ability to name PVCs, what is the benefit of one way vs. the other?

Thanks for your assistance.


Re: Question on ATM interface vs. per vc queueing

Per-VC queuing is usually more efficient.

The physical aspects are essentially the same, it's more an issue of how the process handles the administration of multiple VCs and the resulting traffic.

Without per-vc queuing, the behavior more like a fifo queue, with per-vc queuing, it's more like fair-weighting.

According to articles found on a Cisco-site search: "Per-VC queuing is to ensure that one congested VC does not consume all the memory resources and starve other VCs." In other words, if you have one pool of memory resources for several VCs, and one VC is using most of the pool, then performance of the other VCs will suffer.

CreatePlease to create content