Page 94 - DCAP403_Operating System
P. 94

Unit 5: Scheduling




          2.   Turnaround time: The interval between the submission of a process and the completion  Notes
               of its execution, including the actual running time, plus the time spent sleeping before
               being dispatched or while waiting to access various resources. This is the appropriate
               responsiveness measure for batch production, as well as for time-shared systems that
               maintain multiple batch queues, sharing CPU time among them.

          3.   Meeting deadlines: The ability of the OS to meet pre-defined deadlines for job completion.
               It makes sense only when the minimal execution time of an application can be accurately
               predicted.
          4.   Predictability: The ability of the system to ensure that a given task is executed within a
               certain time interval, and/or to ensure that a certain constant response time is granted
               within a strict tolerance, no matter what the machine load is.

          When the overall system performance is considered, additional scheduling criteria must be taken
          into account:
          1.   Throughput: The rate of completion of processes (processes completed per unit time). This
               is a ‘‘raw’’ measure of how much work is performed, since it depends on the execution
               length of processes, but it’s obviously affected by the scheduling policy.
          2.   User processor utilisation: Time (percentage of unit time) during which the CPU is running
               user processes. This is a measure of how well the system can serve the payload and keep at
               minimum time spent in housekeeping chores.

          3.   Overall processor utilisation: Time percentage during which the CPU is busy. It’s a
               significant criterion for expensive hardware, that must be kept busy as much as possible in

               order to be justify its cost (e.g. supercomputers for numerical calculus applications).
          4.   Resource utilisation balance: It extends the idea of processor utilisation to take into account
               all system resources. A good scheduler should try to keep all the hardware resources in use
               at any time.


               New State                Ready Queue




                                                                          CPU

                                        Suspended
                                        Ready Queue




                                        Suspended
                                        Block Queue




                                        Block Queue



                        Long Term Scheduling
                        Medium Term Scheduling
                        Short Term Scheduling





                                           LOVELY PROFESSIONAL UNIVERSITY                                    87
   89   90   91   92   93   94   95   96   97   98   99