Uniprocessor Scheduling

Transkript

1 Uniprocessor Scheduling Chapter 9 E&CE 354: Processes 0

2 Processor Scheduling Main OS concern: allocation of resources to processes CPU is a key resource; processor scheduling is concerned with its allocation Scheduling concerns include response time throughput processor utilization Presentation sequence this chapter: uniprocessor scheduling next chapter: real-time scheduling E&CE 354: Processes 1

3 Types of Scheduling Long-term scheduling Decision to add to the pool of processes to be executed Medium-term scheduling Decision to add to the number of processes that are partially or fully in main memory Short-term scheduling Decision as to which ready process will be executed by the processor I/O scheduling Decision as to which process s pending I/O request will be handled by an available I/O device E&CE 354: Processes

4 E&CE 354: Processes 3

5 E&CE 354: Processes 4

6 Long-Term Scheduling Determines which programs are admitted to the system for processing Controls the degree of multiprogramming More processes, smaller percentage of time each process is executed E&CE 354: Processes 5

7 Medium-Term Scheduling Part of the swapping function Based on the need to manage the degree of multiprogramming if virtual memory not used, then memory management also an issue when swapping in, consider memory requirements E&CE 354: Processes 6

8 Short-Term Scheduling Known as the dispatcher Executes most frequently Invoked immediately after processing of an event Clock interrupts I/O interrupts Operating system calls Signals E&CE 354: Processes 7

9 Short-Tem Scheduling Criteria broadly categorized over two dimensions A. user vs system oriented B. performance vs non-performance A. user vs system User-oriented response time; turnaround time System-oriented utilization of the processor E&CE 354: Processes 8

10 Short-Term Scheduling Criteria [] B. performance vs not-performance performance-related quantitative [measurable] response time; throughput not performance related qualitative predictability E&CE 354: Processes 9

11 Scheduling Policies priority based process history/community based first-come, first-served (FCFS) round-robin (RR) shortest process next (SPN) shortest remaining time (SRT) highest response ratio next (HRRN) feedback (FB) fair share scheduling E&CE 354: Processes 10

12 Priorities Each process is assigned execution priority Scheduler will always choose a process of higher priority over one of lower priority Supporting OS data structures multiple ready queues, one per priority level priority queue Drawback: lower-priority may suffer starvation to alleviate, allow OS to change process priority based on its age or execution history E&CE 354: Processes 11

13 E&CE 354: Processes 1

14 Decision Mode Nonpreemptive once a process is in the running state, it will continue until it yields the CPU, terminates or blocks itself for I/O Preemptive currently running process may be interrupted and moved to the Ready state by the operating system generally better responsiveness E&CE 354: Processes 13

15 First-Come-First-Served (FCFS) A B C D E Each process joins the Ready queue Proc A Arival 0 Ser 3 When the current process ceases to execute, the oldest process in the Ready queue is selected B C D E E&CE 354: Processes 14

16 First-Come-First-Served [] A short process may have to wait a very long time before it can execute Favors CPU-bound processes I/O processes have to wait until CPU-bound process completes Definition of turnaround time: the residence time, the total time that the item spends in the system (waiting time plus the service time) E&CE 354: Processes 15

17 Round-Robin (RR) A B C D E Proc Arival Ser Uses preemption based on a clock An amount of time is determined that allows each process to use the processor for that length of time here, time quantum, q = 1 time unit E&CE 354: Processes 16 A B C D E

18 Round-Robin [] Clock interrupt is generated at periodic intervals When an interrupt occurs, the currently running process is placed in the ready queue Next ready job is selected Known as time slicing poor performance for I/O bound processes since compute-bound processes tend receive an unfair portion of processor time E&CE 354: Processes 17

19 Shortest Process Next (SPN) Proc Ariv al Ser A 0 3 B C D E A B C D E Nonpreemptive policy Process with shortest expected processing time is selected next Short process jumps ahead of longer processes E&CE 354: Processes 18

20 Shortest Process Next [] Predictability of longer processes is reduced If estimated time for process not correct, the operating system may abort it Possibility of starvation for longer processes For interactive processes, common technique for predicting future value on the basis of a time series of past values: S n+1 = αt n + (1 - α)s n where: S i is predicted value for i-th instance T i actual processor execution in ith instance 0 α 1 E&CE 354: Processes 19

21 Shortest Remaining Time (SRT) A B C D E Proc Arival Ser A 0 3 Preemptive version of shortest process next policy B C Must estimate processing time D 6 5 E 8 E&CE 354: Processes 0

22 Highest Response Ratio Next (HRRN) A B C D E Choose next process with the greatest ratio Proc Arival Ser A 0 3 time spent waiting + expected service time expected service time B C D 6 5 E 8 E&CE 354: Processes 1

23 E&CE 354: Processes

24 Feedback A B X X X X X X X X X C D E X X X X X X X X X X X Proc Arival Ser Penalize jobs that have been running longer Don t know remaining time process needs to execute A B C D E 8 E&CE 354: Processes 3

25 Fair-Share Scheduling (FSS) Some applications run as a collection of processes (threads) Primary concern is with the performance of the entire application, not just individual processes Scheduling decisions must be based on process sets Basic idea: give each set a fixed (fair) share of CPU give more resources to process sets which have had less than their fair share give less resources to process sets which have had more than their fair share E&CE 354: Processes 4

26 Fair-Share Scheduling [] Scheduling based on priority, where higher value of the priority the lower is the scheduling priority: [interval i, process j, process set/group k] CPU j ( i 1) CPU j ( i) = GCPU k ( i 1) GCPU k ( i) = CPU j ( i) GCPU k ( i) Pj ( i) = Base j W where : CPU j Base W k j j GCPU ( i) k = processor utilization by process P ( i) = priority of process = base priority of process = weighting assigned to group k, and 0 W j k ( i) = processor utilization of group k through interval i j at start of interval i j thru interval i k 1and Wk = 1 E&CE 354: Processes 5 k

27 Assumption: w1 = w = 0.5 E&CE 354: Processes 6

28 Traditional UNIX Scheduling Multilevel feedback using round robin within each of the priority queues Each process has a base priority If process does not complete within 1 second, it is preempted Priorities are recomputed once per second Base priority divides all processes into fixed bands of priority levels Adjustment factor sets the relative priority of a process within its assigned band E&CE 354: Processes 7

29 Bands Decreasing order of priority Swapper Block I/O device control File manipulation Character I/O device control User processes Bias towards I/O utilization E&CE 354: Processes 8

30 Priority Priority based on process type and execution history: CPU j P ( i) j ( i) = = CPU j ( i 1) CPU Base j + j ( i) + nice j CPU j (i) = processor utilization by process j thru interval i P j (i) = priority of process j at start of interval i Base j = base priority of process j nice j = user-controllable adjustment factor E&CE 354: Processes 9

31 nice parameter not used E&CE 354: Processes 30