Scheduling

Alternating Sequence of CPU And I/O Bursts

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.

Histogram of CPU-burst Times

Computing Environments

Batch E.g., supercomputing centers,

mainframes/workstations for business computing.

Job Owner Job Name Queue State #Nds Time Used% Time

Max Time

427130

kensong A3i dque ON HOLD 24 0% 24:00:00

427226

kensong A3i dque ON HOLD 24 0% 24:00:00

460157

jmehring SCEC_CyberShake_PAS_2 dque Queued 144 0% 24:00:00

460233

stolbov cocuphGgga long RUNNING 8 02:42:53 3% 96:00:00

460234

stolbov cocuphGgga long ON HOLD 8 0% 96:00:00

462375

stolbov cocuphMgga long ON HOLD 8 0% 96:00:00

462376

stolbov cocuphMgga long ON HOLD 8 0% 96:00:00

464535

zhaol SCEC_LAB_TOMO dque RUNNING 256 02:05:07 9% 24:00:00

470342

yujiewu q192a_1 dque RUNNING 64 02:43:06 11% 24:00:00

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

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.

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.

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

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).

Optimization Criteria

Max CPU utilization Max throughput Min turnaround time Min waiting time Min response time

Optimization Criteria: Conflicting

Maximize throughput Execute all shortest jobs first.

Minimize turnaround time Turnaround time is increased significantly if long jobs

are never executed.

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.

Real-time systems: Meeting deadlines

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!

First-Come, First-Served (FCFS) Scheduling

Process Burst Time

P1 24

P2 3

P3 3

Suppose that the processes arrive in the order: P1 , P2 , P3

The Gantt Chart for the schedule is:

Waiting time for P1 = 0; P2 = 24; P3 = 27 Average waiting time: (0 + 24 + 27)/3 = 17

P1 P2 P3

24 27 300

FCFS Scheduling (Cont.)

Suppose that the processes arrive in the order P2 , P3 , P1 .

The Gantt chart for the schedule is:

Waiting time for P1 = 6; P2 = 0; P3 = 3 Average waiting time: (6 + 0 + 3)/3 = 3 Much better than previous case.

P1P3P2

63 300

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.

P0 P1 P2 P3

0 19 20 21 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

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

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 Idle I/O Devices Idle

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.

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.

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.

Process Arrival Time Burst Time

P1 0.0 7

P2 2.0 4

P3 4.0 1

P4 5.0 4

Example of Non-Preemptive SJF

P1 P3 P2

73 160

P4

8 12

Average waiting time = (0 + 6 + 3 + 7)/4 - 4

Example of Non-Preemptive SJF

P1 P3 P2

73 160

P4

8 12

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.

Preemptive SJF

Process Arrival Time Burst TimeP1 0.0 7

P2 2.0 4

P3 4.0 1

P4 5.0 4

P1

42 110 5 7 16

Process Arrival Time Burst TimeP1 0.0 7

P2 2.0 4

P3 4.0 1

P4 5.0 4

Process Arrival Time Burst Time

P1 0.0 7

P2 2.0 4

P2 Remainder = 4.P1 Remainder = 5. Result: P1 preempted at time 2.

P1 P3P2

42 110

P4

5 7

P2 P1

16

Process Arrival Time Burst Time

P1 0.0 7

P2 2.0 4

P3 4.0 1

P1: 5

P2: 2 P2 Preempted. P3 completes at time 5.P3: 1P1 P3P2

42 110

P4

5 7

P2 P1

16

Process Arrival Time Burst Time

P1 0.0 7

P2 2.0 4

P4 5.0 4

P1: 5

P2: 2P4: 4

P1 P3P2

42 110

P4

5 7

P2 P1

16

P2 Completes at time 7.

P1: Remaining time of 5.

P4: Remaining time of 4.

P1 P3P2

42 110

P4

5 7

P2 P1

16

Process Arrival Time Burst Time P1 0.0 7 P2 2.0 4 P3 4.0 1 P4 5.0 4

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.

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.

Homework #2

Please answer the following question from Chapter 5.

5.2, 5.4, 5.5, 5.6, 5.7, 5.8, 5.10.

Scheduling in Batch Systems

Three level scheduling

Memory Scheduler

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

Admission Scheduler

Based on “degree of multiprogramming” Process mix.

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.

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.

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.

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.

Scheduling in Interactive Systems

Round Robin Scheduling a) list of runnable processes b) list of runnable processes after B uses up its quantum

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.

Time Quantum and Context Switch Time

Multilevel Queue Scheduling

Multilevel Queue

Ready queue is partitioned into separate queues (e.g.,):

foreground (interactive) background (batch)

Each queue has its own scheduling algorithm, foreground – RRbackground – FCFS

Processes do not change queues

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

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

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

Example of Multilevel Feedback Queue

Three queues: Q0 – time quantum 8 milliseconds Q1 – time quantum 16 milliseconds Q2 – FCFS

Example of Multilevel Feedback Queue

Scheduling A new job enters queue Q0 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 Q1.

At Q1 job is again served RR and receives 16 additional milliseconds. If it still does not complete, it is preempted and moved to queue Q2.

Multilevel Feedback Queues