0% found this document useful (0 votes)
9 views21 pages

lecture03e

The document provides an overview of real-time scheduling algorithms, including clock-driven, weighted round-robin, and priority-driven methods. It discusses the characteristics, strengths, and weaknesses of each approach, emphasizing the importance of scheduling in real-time systems. The document also highlights the differences between dynamic and static systems and the validation challenges associated with priority-driven scheduling.

Uploaded by

noejr.setenta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
9 views21 pages

lecture03e

The document provides an overview of real-time scheduling algorithms, including clock-driven, weighted round-robin, and priority-driven methods. It discusses the characteristics, strengths, and weaknesses of each approach, emphasizing the importance of scheduling in real-time systems. The document also highlights the differences between dynamic and static systems and the validation challenges associated with priority-driven scheduling.

Uploaded by

noejr.setenta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

Overview of Real-Time

Scheduling

COE186 | Real-Time Systems Development

Slides from
Dr Colin Perkins
Real-Time and Embedded Systems
University of Glasgow
1
Lecture Outline
• Overview of real-time scheduling algorithms
– Clock-driven
– Weighted round-robin
– Priority-driven
• Dynamic vs. static
• Deadline scheduling: EDF and LST
• Validation
• Outline relative strengths, weaknesses

Material corresponds to chapter 4 of Liu’s book

Overview of Real-Time Scheduling 2


Approaches to Real-Time Scheduling
Different classes of scheduling algorithm used in real-time systems:
• Clock-driven
– Primarily used for hard real-time systems where all properties of all jobs
are known at design time, such that offline scheduling techniques can be
used
• Weighted round-robin
– Primarily used for scheduling real-time traffic in high-speed, switched
networks
• Priority-driven
– Primarily used for more dynamic real-time systems with a mix of time-
based and event-based activities, where the system must adapt to changing
conditions and events

Look at the properties of each in turn…


Overview of Real-Time Scheduling 3
Clock-Driven Scheduling
• Decisions about what jobs execute when are made at specific time
instants
– These instants are chosen before the system begins execution
– Usually regularly spaced, implemented using a periodic timer interrupt
• Scheduler awakes after each interrupt, schedules the job to execute for the next
period, then blocks itself until the next interrupt
• E.g. the helicopter example with an interrupt every 1/180th of a second
• E.g. the furnace control example, with an interrupt every 100ms

• Typically in clock-driven systems:


– All parameters of the real-time jobs are fixed and known
– A schedule of the jobs is computed off-line and is stored for use at run-
time; as a result, scheduling overhead at run-time can be minimized
– Simple and straight-forward, not flexible

[Will discuss in more detail in lecture 4]


Overview of Real-Time Scheduling 4
Weighted Round-Robin Scheduling
• Regular round-robin scheduling is commonly used for scheduling
time-shared applications
– Every job joins a FIFO queue when it is ready for execution
– When the scheduler runs, it schedules the job at the head of the queue to
execute for at most one time slice
• Sometimes called a quantum – typically O(tens of ms)
– If the job has not completed by the end of its quantum, it is preempted and
placed at the end of the queue
– When there are n ready jobs in the queue, each job gets one slice every n
time slices (n time slices is called a round)

– Only limited use in real-time systems

Overview of Real-Time Scheduling 5


Weighted Round-Robin Scheduling
• In weighted round robin each job Ji is assigned a weight wi; the
job will receive wi consecutive time slices each round, and the
duration of a round is ∑i=1 wi
n

– Equivalent to regular round robin if all weights equal 1


– Simple to implement, since it doesn’t require a sorted priority queue

• Partitions capacity between jobs according to some ratio


• Offers throughput guarantees
– Each job makes a certain amount of progress each round

Overview of Real-Time Scheduling 6


Weighted Round-Robin Scheduling
• By giving each job a fixed fraction of the processor time, a round-
robin scheduler may delay the completion of every job
– A precedence constrained job may be assigned processor time, even while
it waits for its predecessor to complete; a job can’t take the time assigned
to its successor to finish earlier
– Not an issue for jobs that can incrementally consume output from their
predecessor, since they execute concurrently in a pipelined fashion
• E.g. Jobs communicating using Unix pipes
• E.g. Wormhole switching networks, where message transmission is carried out
in a pipeline fashion and a downstream switch can begin to transmit an earlier
portion of a message, without having to wait for the arrival of the later portion

• Weighted round-robin is primarily used for real-time networking;


will discuss more in lecture 16
Overview of Real-Time Scheduling 7
Priority-Driven Scheduling
• Assign priorities to jobs, based on some algorithm
• Make scheduling decisions based on the priorities, when events
such as releases and job completions occur
– Priority scheduling algorithms are event-driven
– Jobs are placed in one or more queues; at each event, the ready job with the
highest priority is executed
– The assignment of jobs to priority queues, along with rules such a whether
preemption is allowed, completely defines a priority scheduling algorithm

• Priority-driven algorithms make locally optimal decisions about


which job to run
– Locally optimal scheduling decisions are often not globally optimal
– Priority-driven algorithms never intentionally leave any resource idle
• Leaving a resource idle is not locally optimal

Overview of Real-Time Scheduling 8


Example: Priority-Driven Scheduling
• Consider the following task:
– Jobs J1, J2, …, J8, where Ji had higher priority than Jk if i < k
J1 0/3
Release time

J2 0/1 J3 0/2 J4 0/2


Execution time

J5 4/2 J6 0/4

J7 0/4 J8 0/1

– Jobs are scheduled on two processors P1 and P2


– Jobs communicate via shared memory, so communication cost is negligible
– The schedulers keep one common priority queue of ready jobs
– All jobs are preemptable; scheduling decisions are made whenever some
job becomes ready for execution or a job completes
Overview of Real-Time Scheduling 9
Example: Priority-Driven Scheduling
Time Not yet Released but not Ready to run P1 P2 Completed
released yet ready to run
0
1
2
3
4
5
6
7
8
J1 0/3 Release time
9
10 J2 0/1 J3 0/2 J4 0/2

11 J5 4/2 J6 0/4
Execution time

12
J7 0/4 J8 0/1

Overview of Real-Time Scheduling 10


Example: Priority-Driven Scheduling
• Note: The ability to preempt lower priority jobs slowed down the
overall completion of the task
– This is not a general rule, but shows that priority scheduling results can be
non-intuitive
– Different priority scheduling algorithms can have very different properties

• Tracing execution of jobs using tables is an effective way to


demonstrate correctness for systems with periodic tasks and
fixed timing constraints, execution times, resource usage
– Show that the system enters a repeating pattern of execution, and each
hyper-period of that pattern meets all deadlines
– Proof by exhaustive simulation
• Provided the system has a manageably small number of jobs

Overview of Real-Time Scheduling 11


Priority-Driven Scheduling
• Most scheduling algorithms used in non real-time systems are
priority-driven
– First-In-First-Out
Assign priority based on release time
– Last-In-First-Out
– Shortest-Execution-Time-First
Assign priority based on execution time
– Longest-Execution-Time-First

• Real-time priority scheduling assigns priorities based on deadline


or some other timing constraint:
– Earliest deadline first
– Least slack time first
– Etc.

Overview of Real-Time Scheduling 12


Priority Scheduling Based on Deadlines
• Earliest deadline first (EDF)
– Assign priority to jobs based on deadline
– Earlier the deadline, higher the priority
– Simple, just requires knowledge of deadlines

• Least Slack Time first (LST)


– A job Ji has deadline di, execution time ei, and was released at time ri
– At time t < di:
• Remaining execution time trem = ei - (t - ri)
• Slack time tslack = di - t - trem
– Assign priority to jobs based on slack time, tslack
– The smaller the slack time, the higher the priority
– More complex, requires knowledge of execution times and deadlines
• Knowing the actual execution time is often difficult a priori, since it depends
on the data, need to use worst case estimates (⇒ poor performance)
Overview of Real-Time Scheduling 13
Optimality of EDF and LST
• These algorithms are optimal
– i.e. they will always produce a feasible schedule if one exists
– Constraints: on a single processor, as long as preemption is allowed and
jobs do not contend for resources

Overview of Real-Time Scheduling 14


Optimality of EDF and LST: Proof
1. Any feasible schedule can be transformed into an EDF schedule
• If Ji is scheduled to execute before Jk, but Ji’s deadline is later than Jk’s either:
• The release time of Jk is after the Ji completes ⇒ they’re already in EDF order
• The release time of Jk is before the end of the interval in which Ji executes:

Ji Jk

rk dk di
• Swap Ji and Jk (this is always possible, since Ji’s deadline is later than Jk’s)
Jk Jk Ji

• Move any jobs following idle periods forward into the idle period

Jk Jk Ji

⇒ the result is an EDF schedule


2. So, if EDF fails to produce a feasible schedule, no feasible schedule exists
• If a feasible schedule existed it could be transformed into an EDF schedule,
contradicting the statement that EDF failed to produce a feasible schedule
[Proof for LST is similar]
Overview of Real-Time Scheduling 15
Non-Optimality of EDF and LST
• Neither algorithm is optimal if jobs are non-preemptable or if
there is more than one processor
– The book has examples which demonstrate EDF and LST producing
infeasible schedules in these cases
• This includes non-strict LST scheduling
– Proof-by-counterexample

• EDF and LST are simple priority-driven scheduling algorithms;


introduced to show how we can reason about such algorithms
– Lectures 5-8 discuss other priority-driven scheduling algorithms

Overview of Real-Time Scheduling 16


Dynamic vs. Static Systems
• If jobs are scheduled on multiple processors, and a job can be
dispatched from the priority run queue to any of the
processors, the system is dynamic
• A job migrates if it starts execution on one processor and is
resumed on a different processor
• If jobs are partitioned into subsystems, and each subsystem is
bound statically to a processor, we have a static system
• Expect static systems to have inferior performance (in terms of
overall response time of the jobs) relative to dynamic systems
– But it is possible to validate static systems, whereas this is not always true
for dynamic systems
– For this reason, most hard real time systems are static

Overview of Real-Time Scheduling 17


Effective Release Times and Deadlines
• Sometimes the release time of a job may be later than that of its
successors, or its deadline may be earlier than that specified for
its predecessors
• This makes no sense: derive an effective release time or effective
deadline consistent with all precedence constraints, and schedule
using that
– Effective release time
• If a job has no predecessors, its effective release time is its release time
• If it has predecessors, its effective release time is the maximum of its release
time and the effective release times of its predecessors
– Effective deadline
• If a job has no successors, its effective deadline is its deadline
• It if has successors, its effective deadline is the minimum of its deadline and
the effective deadline of its successors

Overview of Real-Time Scheduling 18


Effective Release Times and Deadlines

• Note: definition of effective deadline does not take


into account execution time of successor jobs
– Would be more accurate, and needs to be done on multiprocessor systems
– But: unnecessary on single processor with preemptable jobs

– Feasible to schedule any set of jobs according to their actual release times
and deadline, iff feasible to schedule according to effective release times
and deadlines
• Ignore precedence constraints, schedule using effective release times and
deadlines as if all jobs independent
• Resulting schedule might meet deadlines but not precedence constraints
• – If so, always possible to swap order of jobs within the schedule to meet
deadlines and precedence constraints

Overview of Real-Time Scheduling 19


Validating Priority-Driven Scheduling
• Priority-driven scheduling has many advantages over clock-
driven scheduling
– Better suited to applications with varying time and resource requirements,
since needs less a priori information
– Run-time overheads are small
• But not widely used until recently, since difficult to validate
– Scheduling anomalies can occur for multiprocessor or non-preemptable
systems, or those which share resources
• Reducing the execution time of a job in a task can increase the total response
time of the task (see book for example)
• Not sufficient to show correctness with worse-case execution times, need to
simulate with all possible execution times for all jobs comprising a task
– Can be proved that anomalies do not occur for independent, preemptable,
jobs with fixed release times executed using any priority-driven scheduler
on a single processor
• Various stronger results exist for particular priority-driven algorithms
Overview of Real-Time Scheduling 20
Summary
• Have outlined different approaches to scheduling:
– Clock-driven
– Weighted round-robin
– Priority-driven
and some of their constraints

• Next session will be a tutorial to review the material covered to


date, before we move onto detailed discussion of scheduling

Overview of Real-Time Scheduling 21

You might also like