This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: CS-350: Fundamentals of Computing Systems Page 1 of 4Lecture Notes Azer Bestavros. All rights reserved. Reproduction or copying (electronic or otherwise) is expressly forbidden except for students enrolled in CS-350. General Processor Sharing From non-preemptive FCFS to preemptive round-robin scheduling The queuing systems (and networks) we have studied so far helped us model how processes consume resources and how the overall performance of the system depends on conflicts over these resources (reflected by queuing delays, etc.). One assumption we made all along is that the use of a resource (i.e. the queue server) is exclusiveonly one process is allowed to be using the CPU at any point in time. In reality, though, it may be possible for the same resource to be shared by multiple processes. In a sense, each process would get a virtual resource whose capacity is 1/n times the capacity of the original resource, where n is the number of processes sharing that resource. Notice that such a model is different from any model we have considered so far in that the server capacity is a function of the number of processes in the system. Also notice that for such a resource, there could be no queuing delays since no process will ever have to wait for the resource! Rather than wait for the resource, the effect of higher load is a reduction in the speed (or service rate) of the server. To further understand GPS, it is helpful to contrast it (and approximate it) to queuing models/networks we are aware of. We do so next. Round Robin as an approximation of GPS Round Robin (RR) is a well-known mechanism that allows a resource (e.g. processor) to be shared by a number of processes by allocating to each one of these processes the processor for a finite amount of timec(the time slice or quantum), after which the process must return to the ready queue and wait for another turn. We can model a RR system by the queuing network shown in Figure 1....
View Full Document
- Spring '09