Scheduling and

Scheduling Algorithms
IT6 – Operating Systems
Scheduling
 refers to the way processes are assigned priorities in
a priority queue.
 This assignment is carried out by software known as
a scheduler.
 Scheduler decides the process to run first by using a
scheduling algorithm
 Long-term scheduler (or job scheduler) – selects which
processes should be brought into the ready queue
 Short-term scheduler (or CPU scheduler) – selects
which process should be executed next and allocates CPU
2
Scheduling Algorithms
 the method by which processes are given access to
system resources, usually processor time.
 The need for a scheduling algorithm arises from the
requirement of most modern systems to execute more
than one process at a time
 Scheduling algorithms are generally only used in a time
slice multiplexing kernel - in order to effectively load
balance a system, the kernel must be able to suspend
execution of threads forcibly in order to begin
execution of the next thread.
3
Performance Metrics
 CPU Efficiency
 Throughput
 Turnaround time
 Waiting Time
 Response Time
 Fairness

4
CPU Efficiency
 Sometimes referred to also as CPU utilization
 the average percentage of the hardware (or the
CPU) that is actually used.
 If the utilization is high, you are getting more
value for the money invested in buying the
computer.
 keeping CPU busy 100% of the time
 Minimize idle times for the CPU
5
Throughput
 the number of jobs completed in a unit of
time.
 If there are more jobs completed, there should
be more happy users.
 Maximize the number of jobs processed per
hour.

6
Turnaround Time
 the average time from submitting a job until it
terminates.
 This is the sum of the time spent waiting in
the queue, and the time actually running.

7
Waiting Time
 the time a job waits (in the ready state) until it
runs
 reducing the time a job waits until it runs also
reduces its response time
 As the system has direct control over the
waiting time, but little control over the actual
run time, it should focus on the wait time

8
Response Time
 Minimize response time by quickly dealing
with interactive requests, and letting batch
requests wait.
 This normalizes all jobs to the same scale:
long jobs can wait more, and don’t count
more than short ones.

9
Preemptive Scheduling
 A process scheduling strategy in which the
processing of a job is interrupted and the CPU
is transferred to another job. Also called
context switching.
 Temporarily suspend the logically runnable
processes

10
Non-Preemptive Scheduling
 A job scheduling strategy in which the job
captures the processor and begins execution
and runs uninterrupted until it issues an IO
request or it is finished.
 Run a process to completion

11
Non-preemptive Scheduling Algo.
 First-come, first-served (FCFS)
 Shortest Job Next (SJN)
 Priority Scheduling

12
First-Come First-Served (FCFS)
 a service policy whereby the requests of
customers or clients are attended to in the
order that they arrived, without other biases or
preferences.
 behavior: what comes in first is handled first,
what comes in next waits until the first is
finished, etc.

13
Shortest Job Next (SJN)
 Associate the length of the next CPU burst with each
process
 The job which has the shortest burst time gets the
CPU
 Assign the process with shortest CPU burst
requirement to the CPU
 Specially suitable to batch processing (long term
scheduling)

14
Priority Scheduling
 Most common method in batch systems, but may
give slow turnaround to some users.
 Gives preferential treatment to important jobs.
 Jobs with the same priority are treated FCFS.
 Priority could be based on:
 Least amount of memory required
 Least number of peripheral devices needed
 Shorted estimated CPU time
 Time already spent waiting.
15
Priority Function
 Memory requirements
 Important due to swapping overhead
 Smaller memory size => Less swapping overhead
 Smaller memory size => More processes can be
serviced
 Timeliness
 Dependent upon the urgency of a task
 Deadlines

16
Priority Function
 Total service time
 Total CPU time consumed by the process during
its lifetime
 Equals attained service time when the process
terminates
 Higher priority for shorter processes
 Preferential treatment of shorter processes reduces
the average time a process spends in the system

17
Priority Function
 External priorities
 Differentiate between classes of user and system
processes
 Interactive processes => Higher priority
 Batch processes => Lower priority
 Accounting for the resource utilization
 Attained service time
 Total time when the process is in the running state

18
Shortest Remaining Time (SRT)
 Preemptive version of shortest job next
scheduling
 Preemptive in nature (only at arrival time)
 The processor is allocated to the job closest to
completion.
 Can’t be implemented on interactive system
because it requires advance knowledge of
required CPU time.
19
Round Robin (RR)
 Preemptive in nature
 Preemption based on time slices or time quanta;
Time quantum between 10 and 100 milliseconds
 All user processes treated to be at the same priority
 No process is allocated CPU for more than 1
quantum in a row
 Timer interrupt results in context switch and the
process is put at the rear of the ready queue

20
Round Robin (RR)
 Ready queue treated as a circular queue
 New processes added to the rear of the ready queue
 Preempted processes added to the rear of the ready
queue
 Scheduler picks up a process from the head of the
queue and dispatches it with a timer interrupt set after
the time quantum
 CPU burst < 1 quantum ) process releases CPU
voluntarily
21
Round Robin (RR)
 Large time quantum => FIFO scheduling
 Small time quantum => Large context
switching overhead
 Rule of thumb: 80% of the CPU bursts
should be shorter than the time quantum

22
Comparison of scheduling methods
 Average waiting time
 Average turnaround time

23
Summary
 Scheduling - which is what “resource
management” is in the context of processes -
is a hard problem.
 It requires detailed knowledge, e.g. how long
a job will run, which is typically not
available. And then it turns out to be NP-
complete.

24
Summary
 However, this doesn’t mean that operating
systems can’t do anything. The main idea is
to use preemption.
 This allows the operating system to learn
about the behavior of different jobs, and to
reconsider its decisions periodically.

25
Summary
 Periodical preemptions are not guaranteed to
improve average response times, but they do.
 The reason is that the distribution of process
runtimes is heavy-tailed.
 This is one of the best examples of a
widespread operating system policy that is
based on an empirical observation about
workloads.
26
END OF LECTURE

27