1. Scheduling

2. Alternating Sequence of CPU And I/O Bursts

3. Simple Categories of Processes CPU-bound process is one that has more and larger CPU bursts (spends most of its time computing). An I/O-bound process spends most of its time waiting for I/O.

4. Histogram of CPU-burst Times

5. Computing Environments Batch E.g., supercomputing centers, mainframes/workstations for business computing.

6. JobOwnerJob NameQueueState#NdsTime Used % Time Max Time 42713 0 kensongA3idqueON HOLD240%24:00:00 42722 6 kensongA3idqueON HOLD240%24:00:00 46015 7 jmehringSCEC_CyberShake_PAS_2dqueQueued1440%24:00:00 46023 3 stolbovcocuphGggalongRUNNING802:42:533%96:00:00 46023 4 stolbovcocuphGggalongON HOLD80%96:00:00 46237 5 stolbovcocuphMggalongON HOLD80%96:00:00 46237 6 stolbovcocuphMggalongON HOLD80%96:00:00 46453 5 zhaolSCEC_LAB_TOMOdqueRUNNING25602:05:079%24:00:00 47034 2 yujiewuq192a_1dqueRUNNING6402:43:0611%24:00:00

7. Interactive Systems (e.g., Gandalf). Real-Time systems.

8. CPU Scheduler Selects from among the processes in memory (on ready queue), and allocates the CPU to one of them. CPU scheduling decisions may take place when a process: 1.Switches from running to waiting state. 2.Switches from running to ready state. 3.Switches from waiting to ready. 4.Terminates. Scheduling under 1 and 4 is nonpreemptive. All other scheduling is preemptive.

9. Dispatcher Dispatcher module gives control of the CPU to the process selected by the short-term scheduler; this involves: switching context switching to user mode jumping to the proper location in the user program to restart that program Dispatch latency – time it takes for the dispatcher to stop one process and start another running.

10. Possible Scheduling Goals CPU utilization – keep the CPU as busy as possible Throughput – # of processes that complete their execution per time unit Turnaround time – amount of time to execute a particular process

11. Possible Scheduling Goals (continued) Waiting time – amount of time a process has been waiting in the ready queue Response time – amount of time it takes from issuing a command and getting response (interactive systems, PCs).

12. Optimization Criteria Max CPU utilization Max throughput Min turnaround time Min waiting time Min response time

13. Optimization Criteria: Conflicting Maximize throughput Execute all shortest jobs first. Minimize turnaround time Turnaround time is increased significantly if long jobs are never executed.

14. Goals of Scheduling Algorithms All systems: Fairness Balance (keep all parts of the system busy). Batch Systems: Maximize throughput. Minimize turnaround time CPU utilization. Interactive systems: Response time.

15. Real-time systems: Meeting deadlines

16. First-Come, First-Served (FCFS) Scheduling Allocate CPU to processes based on arrival order. Non-preemptive. Process executes until completes or blocks on I/O or other system resource. Used in batch systems (with some modifications). Not a good idea for a timesharing system!

17. First-Come, First-Served (FCFS) Scheduling ProcessBurst Time P 1 24 P 2 3 P 3 3 Suppose that the processes arrive in the order: P 1, P 2, P 3

18. The Gantt Chart for the schedule is: Waiting time for P 1 = 0; P 2 = 24; P 3 = 27 Average waiting time: (0 + 24 + 27)/3 = 17 P1P1 P2P2 P3P3 2427300

19. FCFS Scheduling (Cont.) Suppose that the processes arrive in the order P 2, P 3, P 1. The Gantt chart for the schedule is: Waiting time for P 1 = 6; P 2 = 0 ; P 3 = 3 Average waiting time: (6 + 0 + 3)/3 = 3 Much better than previous case. P1P1 P3P3 P2P2 63300

20. One problem with FCFS is that average waiting time can be quite large. Another problem is the Convoy effect. Consider: 4 processes; 1 CPU-bound with CPU burst of 20 units followed by an I/O request requiring 10 units. 3 I/O bound processes with 1 unit of CPU burst followed by 10 units of I/O.

21. P0 P1 P2 P3 0 19 20 21 22

22. P0 P1 P2 P3 0 19 20 21 22 I/O 1 I/O 2 I/O 3 I/O 4 P0: 7 P1: 8 P2: 9 P3: 10

23. P0 P1 P2 P3 P0 0 19 20 21 22 29 I/O 1 I/O 2 I/O 3 I/O 4 P1: 1 P2: 2 P3: 3 I/O Devices Idle CPU Idle

24. P0 P1 P2 P3 P0 0 19 20 21 22 29 32 49 I/O 1 I/O 2 I/O 3 I/O 4 I/O Devices Idle CPU IdleI/O Devices Idle

25. FCFS Scheduling (Cont.) Convoy effect short I/O bound processes behind long CPU- bound process. CPU-bound process executes, I/O processes wait in Ready Queue. CPU-bound process completes execution burst and waits on I/O device. IO-Bound processes quickly complete CPU burst and block on I/O device. CPU is idle until I/O completed for CPU-bound process. CPU-bound process resumes, I/O-bound processes complete I/O request and move to RQ. I/O devices idle while CPU-bound process monopolizes CPU.

26. Shortest-Job-First (SJR) Scheduling Associate with each process the length of its next CPU burst. Use these lengths to schedule the process with the shortest time. Two schemes: nonpreemptive – once CPU given to the process it cannot be preempted until completes its CPU burst.

27. Shortest-Job-First (SJR) Scheduling preemptive – if a new process arrives with CPU burst length less than remaining time of current executing process, preempt. This scheme is know as the Shortest-Remaining-Time-First (SRTF). SJF is optimal – gives minimum average waiting time for a given set of processes.

28. ProcessArrival TimeBurst Time P 1 0.07 P 2 2.04 P 3 4.01 P 4 5.04 Example of Non-Preemptive SJF P1P1 P3P3 P2P2 73160 P4P4 812

29. Average waiting time = (0 + 6 + 3 + 7)/4 - 4 Example of Non-Preemptive SJF P1P1 P3P3 P2P2 73160 P4P4 812

30. Preemptive Shortest Job First If a job arrives at the queue with a burst time less than that of the running process, the running process is preempted. Decision only made when a new process enters the queue.

31. Preemptive SJF Process Arrival Time Burst Time P 1 0.07 P 2 2.04 P 3 4.01 P 4 5.04

32. P1P1 42 11 0 57 16 Process Arrival Time Burst Time P 1 0.07 P 2 2.04 P 3 4.01 P 4 5.04

33. ProcessArrival TimeBurst Time P 1 0.07 P 2 2.04 P2 Remainder = 4. P1 Remainder = 5. Result: P1 preempted at time 2. P1P1 P3P3 P2P2 42 11 0 P4P4 57 P2P2 P1P1 16

34. ProcessArrival TimeBurst Time P 1 0.07 P 2 2.04 P 3 4.01 P1: 5 P2: 2P2 Preempted. P3 completes at time 5. P3: 1 P1P1 P3P3 P2P2 42 11 0 P4P4 57 P2P2 P1P1 16

35. ProcessArrival TimeBurst Time P 1 0.07 P 2 2.04 P 4 5.04 P1: 5 P2: 2 P4: 4

36. P1P1 P3P3 P2P2 42 11 0 P4P4 57 P2P2 P1P1 16 P2 Completes at time 7. P1: Remaining time of 5. P4: Remaining time of 4.

37. P1P1 P3P3 P2P2 42 11 0 P4P4 57 P2P2 P1P1 16 Process Arrival Time Burst Time P10.07 P22.04 P34.01 P45.04 P1: Waits from time 2 to time 11 = 9 P2: Waits from time 4 to time 5 = 1 P3: No waiting = 0 P4: Waits from time 5 to time 7 = 2 Average wait = 12/4 = 3.

38. Determining Length of Next CPU Burst Can only estimate the length. Estimate made based on some sort of statistic of historical behavior. Assume BL2 == BL1 (Next same as last). Take mean of last n burst lengths. Exponential average of previous bursts.

39. Homework #2 Please answer the following question from Chapter 5. 5.2, 5.4, 5.5, 5.6, 5.7, 5.8, 5.10.

40. Scheduling in Batch Systems Three level scheduling

41. Memory Scheduler Decisions based on for example: Time since swapped out. Amount of CPU time allocated so far. How large. How important.

42. Admission Scheduler Based on “degree of multiprogramming” Process mix.

43. Priority Scheduling A priority number (integer) is associated with each process The CPU is allocated to the process with the highest priority (smallest integer  highest priority). Preemptive nonpreemptive SJF is a priority scheduling where priority is the predicted next CPU burst time. Problem  Starvation – low priority processes may never execute.

44. Priority Scheduling Solution  Aging – as time progresses increase the priority of the process. Unix has mechanism for user to lower their priority through the nice system call.

45. Scheduling for Interactive Systems: Round Robin (RR) Each process gets a small unit of CPU time (time quantum), usually 10-100 milliseconds. After this time has elapsed, the process is preempted and added to the end of the ready queue. If there are n processes in the ready queue and the time quantum is q, then each process gets 1/n of the CPU time in chunks of at most q time units at once. No process waits more than (n-1)q time units.

46. Scheduling for Interactive Systems: Round Robin (RR) Performance q large  FIFO q small  High overhead: Must be large with respect to context switch, otherwise overhead is too high.

47. Scheduling in Interactive Systems Round Robin Scheduling a) list of runnable processes b) list of runnable processes after B uses up its quantum

48. How long should the quantum be? quantum too short Assume switch time = 5ms and quantum = 20ms: Wasted time = 5/(5+20) = 20% quantum too long: e.g., switch time = 5ms, quantum = 200ms: Wasted time = 5/(5+200) = approx. 2% but if have 100 processes, response time for 200th is pretty bad. This is the quantum Linux uses.

49. Time Quantum and Context Switch Time

50. Multilevel Queue Scheduling

51. Multilevel Queue Ready queue is partitioned into separate queues (e.g.,): foreground (interactive) background (batch) Each queue has its own scheduling algorithm, foreground – RR background – FCFS Processes do not change queues

52. Multilevel Queue Scheduling must also be done between the queues. Fixed priority scheduling; (i.e., serve all from foreground then from background). Possibility of starvation. Time slice – each queue gets a certain amount of CPU time which it can schedule amongst its processes; i.e., 80% to foreground in RR 20% to background in FCFS

53. Multilevel Feedback Queue A process can move between the various queues; aging can be implemented this way. Higher priority queues can preempt lower priority queues. Multilevel-feedback-queue scheduler defined by the following parameters: number of queues scheduling algorithms for each queue

54. Multilevel Feedback Queue method used to determine when to upgrade a process method used to determine when to demote a process method used to determine which queue a process will enter when that process needs service

55. Example of Multilevel Feedback Queue Three queues: Q 0 – time quantum 8 milliseconds Q 1 – time quantum 16 milliseconds Q 2 – FCFS

56. Example of Multilevel Feedback Queue Scheduling A new job enters queue Q 0 which is served RR. When it gains CPU, job receives 8 milliseconds. If it does not finish in 8 milliseconds, job is moved to queue Q 1. At Q 1 job is again served RR and receives 16 additional milliseconds. If it still does not complete, it is preempted and moved to queue Q 2.

57. Multilevel Feedback Queues