* How service providers use the queuing algorithms found in Cisco routers to implement class-of-service functionality
In a recent newsletter entitled “Beware of MPLS service agreements” we commented that we had reviewed the service-level agreements of some of the major suppliers of MPLS-based services. We pointed out that we were surprised to see how those providers handled the situation in which more traffic was placed into the real-time traffic class than that class was designed to handle.
Ever since that newsletter was published we have been hearing from readers on this topic. One person that we heard from recently was Nathan Wilkes, the principal network architect at Virtela. Nathan pointed out that a lot of the differences in terms of how the service providers deal with service classes depends on which equipment vendor they use. For example, Nathan pointed out that in most cases in which class of service is implemented on a Cisco router and VoIP is part of the traffic mix, then Low-Latency Queuing (LLQ) is implemented.
Nathan explained that Cisco’s LLQ is simply a modification to the Class-Based Weighted Fair Queuing model. Using this queuing algorithm, classes of traffic, usually based on Differentiated Services Code Point (DSCP) groups, are guaranteed a minimum bandwidth per class. The difference with LLQ is that they also introduce a Priority Queue to service latency-sensitive traffic without regard to the Weighted Round Robin servicing of the other queues.
The advantage of using LLQ with a priority queue is that voice packets can be transmitted without being buffered or delayed. One of the disadvantages of this approach is that without some form of policing mechanism, a given priority queue could consume all of the bandwidth. Another disadvantage of this approach is that even if there is not any latency-sensitive traffic, the remaining traffic cannot access the priority queue.
Our next newsletter will look at how service providers use the queuing algorithms found in Juniper routers to implement class of service functionality, and will also look at other key issues that enterprise IT organizations need to be aware of as they evaluate and implement MPLS services.
