Distributed Process Scheduling: 5.1 A System Performance Model

Shuman Guo

CSc 8320, Spring 2007

1

Outline

Overview A System Performance Model Processor Pool and Workstation Queuing

Models References

2

Overview[Randy Chow, 97]

Before execution, processes need to be scheduled and allocated with resources

The objective of scheduling Primary: Enhance overall system performance metrics

Process completion time and processor utilization Secondary: achieve location and performance

transparencies This chapter presents a model for capturing the

effect of communication and system architectures on scheduling.

3

Outline

Overview A System Performance Model Processor Pool and Workstation Queuing

Models References

4

A System Performance Model

We used graph models to describe process communication.

5

Four processes mapped to a two-processor multiple computer system

Process Models

Precedence process model: Represent precedence relationships between

processes Minimize total completion time of task

(computation + communication) Communication process model

Represent the need for communication between processes

6

Process Models cont’d

Optimize the total cost of communication and computation

Disjoint process model Processes can be run independently and

completed in finite time Maximize utilization of processors and

minimize turnaround time of processes

7

System Performance Model

8

Attempt to minimize the total completion time of (makespan) of a set of interacting processes

System Performance Model cont’d

Related parameters OSPT= optimal sequential processing time; CPT= concurrent processing time; OCPTideal =optimal concurrent processing time

on an ideal system; Si =ideal speedup obtained by using a multiple

processor system over the best sequential time Sd = the degradation of the system due to actual

implementation compared to an ideal system

9

System Performance Model (Cont.)

10

Pi: the computation time ofthe concurrent algorithm onnode i

(RP 1)

System Performance Model cont’d

11

(The smaller, the better)

System Performance Model cont’d

RP: Relative processing Shows how much loss of speedup is due to the

substitution of the best sequential algorithm by an algorithm better adapted for concurrent implementation but which may have a greater total processing need

Sd Degradation of parallelism due to algorithm

implementation

12

System Performance Model cont’d

RC: Relative concurrency How far from optimal the usage of the n-processor is RC=1 best use of the processors

: Efficiency Loss is loss of parallelism when implemented on a real machine.

can be decomposed into two terms:

= sched + syst

13

Workload Distribution

Performance can be further improved by workload distribution

Load sharing: static workload distribution Dispatch process to the idle processors statically upon

arrival Corresponding to processor pool model

Load balancing: dynamic workload distribution Migrate processes dynamically from heavily loaded

processors to lightly loaded processors Corresponding to migration workstation model

14

Queuing Theory Performance of systems described as queuing

models can be computed using queuing theory. An X/Y/c queue is one where:

X: Arrival Process, Y: Service time distribution, c: Numbers of servers

: arrival rate; : service rate; : migration rate : depends on channel bandwidth, migration protocol,

context and state information of the process being transferred.

15

Processor-Pool and Workstation Queueing Models

16

Static Load SharingDynamic Load Balancing

M for Markovian distribution

Examples of Real World Queuing Systems? [Lawrence]

17

Commercial Queuing SystemsCommercial organizations serving external customersEx. Medical[Huang,07], bank, ATM, gas stations, plumber, garage

Transportation service systemsVehicles are customers or serversEx. Vehicles waiting at toll stations and traffic lights, trucks or ships waiting to be loaded[Yeon,07] ,taxi cabs, fire engines, elevators, buses …

Examples cont’d

Business-internal service systems Customers receiving service are internal to the

organization providing the service Ex. Inspection stations, conveyor belts, computer

support …

Social service systems Ex. Judicial process, the ER at a hospital, waiting lists for

organ transplants or student dorm rooms …

18

References[1] Randy Chow & Theodore Johnson, 1997,“Distributed Operating Systems &

Algorithms”, (Addison-Wesley), p. 149 to 156.[2] Stephen Lawrence.”Queuing & Simulation”. http://209.85.165.104/search?

q=cache:hCreyAHJ8WgJ:leedsfaculty.colorado.edu/lawrence/SYST4060/Lectures/6a%2520%2520Intro%2520to%2520Queueing.ppt+queuing+simulation+ppt+lawrence&hl=en&ct=clnk&cd=1&gl=us

[3] Yeon, Jiyoun; Ko, Byungkon. ” Comparison of Travel Time Estimation Using Analysis and Queuing Theory to Field Data Along Freeways”. Multimedia and Ubiquitous Engineering, 2007. MUE ‘07 International Conference onApril 2007 Page(s):530 - 538 .

[4] Ean-Wen Huang; Der-Ming Liou. ”Performance Analysis of a Medical Record Exchanges Model”. Information Technology in Biomedicine, IEEE Transactions on March 2007 Page(s):153 - 160

19

Thank you!

Any questions?

20