1. Scheduling

2. Content of This Lecture
Why CPU scheduling?
Basic scheduling algorithms
FIFO (FCFS)
Shortest job first
Round Robin
Priority Scheduling
Goals:
Understand how your program is executed on the machine together with other programs

3. Review: State Process Model

4. OS Representation of Process via Process Control Block (PCB)

5. Process Scheduling
Deciding which process/thread should occupy the resource (CPU, disk, etc)
CPU
I want to play
Whose turn is it?
Process 2
Process 3
Process 1

6. Context Switch Switch CPU from one process to another
Performed by scheduler
It includes:
save PCB state of the old process;
load PCB state of the new process;
Flush memory cache;
Change memory mapping (TLB);
Context switch is expensive ( microseconds)
No useful work is done (pure overhead)
Can become a bottleneck
Need hardware support

7. When should the scheduler be called?
A new process is admitted
The running process exits
The running process is blocked
I/O interrupt (some processes will be ready)
Clock interrupt (every 10 milliseconds)

8. Preemptive vs. Non-preemptive scheduling
The running process keeps the CPU until it voluntarily gives up the CPU
process exits
switches to blocked state
Transition 3 is only voluntary
Preemptive scheduling:
The running process can be interrupted and must release the CPU (can be forced to give up CPU) 4
Terminated
Running 1 3
Ready
Blocked

9. Scheduling Objectives
Fairness (equitable shares of CPU)
Priority (highest priority first)
Efficiency (make best use of equipment)
Encouraging good behavior (can’t take advantage of the system)
Support for heavy loads (degrade gracefully)
Adapting to different environments (interactive, real-time, multi-media)

10. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Performance Criteria
Efficiency: keep resources as busy as possible
Throughput: # of processes that complete in unit time
Waiting Time
Total amount of time spent by the process waiting in ready queue
Initial Waiting Time
Amount of time spent by the process waiting in ready queue before it starts executing
Response Time
amount of time from when a job is admitted until it completes
Proportionality:
Assign CPU proportionally to given application weight
Meeting Deadlines:
example, complete by 5pm Dec. 15th 2009!

11. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Process Profiles
I/O – Bound
Does too much I/O to keep CPU busy
CPU – Bound
Does too much computation to keep I/O busy
Process Mix
Scheduling should load balance between I/O bound and CPU-bound processes
Ideal would be to run all equipment at 100% utilization but that would not necessarily be good for response time

12. Simple Processor Scheduling Algorithms
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Simple Processor Scheduling Algorithms
Batch systems
First Come First Serve (FCFS)
Shortest Job First
Interactive Systems
Round Robin
Priority Scheduling …

13. First Come First Serve (FCFS)
Copyright ©: Nahrstedt, Angrave, Abdelzaher
First Come First Serve (FCFS)
Process that requests the CPU FIRST is allocated the CPU FIRST.
Also called FIFO
Used in Batch Systems
Implementation
FIFO queue
A new process enters the tail of the queue
The scheduler selects the process at the head of the queue.
Quiz: Is it Preemptive or Non-preemptive?

14. First Come First Serve (FCFS)
Copyright ©: Nahrstedt, Angrave, Abdelzaher
First Come First Serve (FCFS)
Process that requests the CPU FIRST is allocated the CPU FIRST.
Also called FIFO
Used in Batch Systems
Implementation
FIFO queue
A new process enters the tail of the queue
The scheduler selects the process at the head of the queue.
Quiz: Is it Preemptive or Non-preemptive?
 once a process has CPU, it runs to completion (non-preemptive)!

15. Copyright ©: Nahrstedt, Angrave, Abdelzaher
FCFS Example
Process
Duration
Order
Arrival Time P1 24 1 P2 3 2 P3 4
The final schedule:
P1 (24)
P2 (3)
P3 (4) 24 27
P1 waiting time: 0
P2 waiting time: 24
P3 waiting time: 27
The average waiting time: ( )/3 = 17
Does the execution order affect the average waiting time?

16. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Problems with FCFS
Non-preemptive
Does not minimize AWT
Cannot utilize resources in parallel:
Assume 1 process CPU bounded and many I/O bounded processes
result: Convoy effect, low CPU and I/O Device utilization
Why?

17. Why Convoy Effects with FCFS?
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Why Convoy Effects with FCFS?
Consider n-1 jobs in system that are I/O bound and 1 job that is CPU bound:
I/O bound jobs pass quickly through the ready queue and suspend themselves waiting for I/O.
CPU bound job arrives at head of queue and executes until completion.
I/O bound jobs re-join ready queue and wait for CPU bound job to complete.
I/O devices idle until CPU bound job completes.
When CPU bound job completes, the ready I/O-bounded processes quickly move through the running state and become blocked on I/O events again. CPU becomes idle.

18. Interactive Scheduling Algorithms
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Interactive Scheduling Algorithms
Usually preemptive
Time is sliced into quanta (time intervals)
Scheduling decision is also made at the beginning of each quantum
Performance Criteria
Average response time
Average initial waiting time
Average waiting time
Fairness (or proportional resource allocation)
Representative algorithms:
Round-robin
Priority-based …

19. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Round-robin
One of the oldest, simplest, most commonly used scheduling algorithm
Select process/thread from ready queue in a round-robin fashion (take turns)
Problems:
Does not consider priority
Context switch overhead
Preemption after
quantum
expiration

20. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Round-robin: Example
Process
Duration
Order
Arrival Time P1 3 1 P2 4 2 P3
Suppose time quantum is: 1 unit, P1, P2 & P3 never block P1 P2 P3 P1 P2 P3 P1 P2 P3 P2 10
P1 waiting time: 4
P2 waiting time: 6
P3 waiting time: 6
The average waiting time (AWT): (4+6+6)/3 = 5.33
Note that Initial Waiting Time of P1 is zero, P2’s IWT is 1, and P3’s IWT is 2

21. Choosing the Time Quantum
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Choosing the Time Quantum
Time slice too large
FIFO behavior
Poor initial waiting time
Time slice too small
Too many context switches (overheads)
Inefficient CPU utilization
Heuristic:
70-80% of jobs block within time-slice
Typical time-slice
ms (depends on job priority)

22. Shortest Job First (SJF)
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Shortest Job First (SJF)
Schedule the job with the shortest computation time first
Scheduling in Batch Systems
Two types:
Non-preemptive
Preemptive
Optimal if all jobs are available simultaneously: Gives the best possible AWT (average waiting time)

23. Non-preemptive SJF: Example
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Non-preemptive SJF: Example
Process
Duration
Order
Arrival Time P1 6 1 P2 8 2 P3 7 3 P4 4
P4 (3)
P1 (6)
P3 (7)
P2 (8) 3 9 16 24
P4 waiting time: 0
P1 waiting time: 3
P3 waiting time: 9
P2 waiting time: 16
The total time is: 24
The average waiting time (AWT): ( )/4 = 7

24. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Comparing to FCFS
Process
Duration
Order
Arrival Time P1 6 1 P2 8 2 P3 7 3 P4 4

25. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Comparing to FCFS
Process
Duration
Order
Arrival Time P1 6 1 P2 8 2 P3 7 3 P4 4
P1 (6)
P2 (8)
P3 (7)
P4 (3) 14 21 24 6
The total time is the same.
The average waiting time (AWT): ( )/4 = 10.25
(comparing to 7)
P1 waiting time: 0
P2 waiting time: 6
P3 waiting time: 14
P4 waiting time: 21

26. Copyright ©: Nahrstedt, Angrave, Abdelzaher
Preemptive SJF
Shortest job runs first.
A job that arrives and is shorter than the running job will preempt it

27. A Problem with Preemptive SJF
Copyright ©: Nahrstedt, Angrave, Abdelzaher
A Problem with Preemptive SJF
Starvation
A job may keep getting preempted by shorter ones
Example
Process A with computation time of 1 hour arrives at time 0
But every 1 minute, a short process with computation time of 2 minutes arrives
Result of SJF: A never gets to run
What’s the difference between starvation and a deadlock?