Benefit to QoS on point-to-point or traditional Frame Relay?

bmoney · ‎05-14-2009

When I think of QoS, I typically picture it being used in conjunction with a WAN technology that supports queuing such as MPLS. The benefit in such a scenario seems clear in that the provider maintains traffic prioritization across the WAN. Having said this, is there any benefit to using QoS on WAN links that do not provide this service (i.e., traditional point-to-point or Frame Relay)?

paolo bevilacqua · ‎05-14-2009

Yes, there is a lot of benefits, for example you can carry a phone conversation or make sure that critical applications enjoy reserved bandwidth.

Note that MPLS, does not provide queuing per se.

bmoney · ‎05-14-2009

Ok, so let's say you are using a fully meshed WAN topology with no provider queuing. You configure QoS queues on all of the edge routers. Voice and other important data are given higher priority as they leave the LAN and enter the WAN cloud, but once in the cloud the provider does not recognize said prioritization. In other words, we can prioritize the traffic, but if the provider ignores the queuing, does this not negate the benefit? Just trying to understand.

paolo bevilacqua · ‎05-14-2009

Not really. For example in FR you have CIR that is a measure of some bandwidth that the provider will deliver all the time. Similarly with ATM. You can configure so the router will not exceed this traffic contract.

Joseph W. Doherty · ‎05-14-2009

". . . is there any benefit to using QoS on WAN links that do not provide this service (i.e., traditional point-to-point or Frame Relay)?"

Often yes, and often very much!

Actually, assuming your devices support it, p-2-p is often one of the easiest places to leverage QoS.

QoS, generally, is about managing congestion to insure certain traffic obtains what it needs to provide a certain "quality of service" (rather than best effort).

With the various cloud technologies, usually the internal cloud portion provides much more bandwidth than your ingress/egress connections and/or the provider will guarantee some bandwidth (which you've purchased) across their cloud. If you logically treat the cloud as p-2-p links (not uncommon with clouds that have [permanent] virtual circuits), QoS is often much as you might do with p-2-p (except for different bandwidth connections to the virtual circuit). With MPLS, it's common to provide a full mesh or multi-point topology. For these, cloud egress QoS can be crucial (besides QoS to the p-2-p link or cloud), and that's where MPLS provided QoS becomes important.