Scheduling in CPU

7/28/2019 Scheduling in CPU

1/32

Computer Systems

Performance Evaluation

CSCI 8710

KraemerFall 2008


2/32

Course Content

This four-hour course address computer systems performanceanalysis. It introduces the main concepts and techniques neededto plan the capacity of computer systems, predict their futureperformance under different configurations, and design newapplications that meet performance requirements. The course is

mainly based on the use of analytic queuing network models ofcomputer systems. These techniques are applied to study theperformance of centralized, distributed, parallel, client/serversystems, Web server and e-commerce site performance. Thecourse provides the students with hands-on experience inperformance evaluation through a project. The concept andapplications of software performance engineering are alsocovered.


3/32

Course Web Site

http://www.cs.uga.edu/~eileen/8710/


4/32

Grading Scheme

Your grade in this course will be calculated as the average of thefollowing: Homeworks -- 10% (A selected subset of the homeworks will be

graded. )

Quizzes -- 5% (One per paper presented.)

Paper presentation -- 10% Midterm Exam -- 25%

Final Exam -- 30%

Course project -- 20%

Letter Grades assigned as follows:

90


5/32

Late Policy

Paper presentations must be ready on the scheduled date. Latesubmissions of projects and homeworks will be accepted with apenalty computed as follows: score = Max_Score *Percentage_Score * (0.90)^(days_late).

A day is a 24 hour period, rounded up to the nearest day.

For example: You turn your project in 3 days late, and received a 95% score

based on the work done. Your recorded score will be:

score = 100 * (0.95) * (0.90)^3

score = 100 * (0.95) * (0.729)

score = 69 Exception

In the case that a solution is distributed, no project submissionswill will be accepted after distribution of the solution


6/32


7/32

Textbook

Performance by Design:

Computer Capacity Planning

by Example,

by Daniel A. Menasce, Virgilio A.F.Almeida, and Lawrence W. Dowdy


8/32


9/32


10/32

QoS Metrics

Response time

Throughput

Availability

Reliability

Security

Scalability Extensibility


11/32


12/32

Throughput

Measured in units of work completed over

time. Its a rate.

I/Os/sec

Page downloads/sec

HTTP requests/sec

Jobs/sec

Transactions per second (tps)


13/32


14/32

Throughput example


15/32

Availability

Fraction of time a system is available

(i.e.,operational).

Service interruptions can damage the reputation

of a company, may endanger lives, and maycause financial disasters.

A system with 99.99% availability over 30 days is

unavailable (1-0.9999) x 30 x 24 x 60 = 4.32

minutes.


16/32

Availability Problems


17/32

Admission Control


18/32

Admission Control

Improved response time

Decreased availability


19/32


20/32

Extensibility

Property of a system to constantly evolve to

meet functional and performance

requirements.

Autonomic computing, self-managing systems,self-healing systems.


21/32

Computer System Lifecycle

Functional requirements: what the system has to do and on what typeof platforms.

Non-functional requirements: how well the system has to accomplish

its functions. Service Level Agreements (SLA) are established.

In many cases, non-functional requirements have been neglected or

considered only at system test time!


22/32


23/32


Components are implemented.- some are new

- some are re-used

- some are adapted

Components are interconnected to form a system

Components should be instrumented as they are built


24/32


25/32


Configuration parameters have to be set in order to meet the SLAs. e.g., TCP parameters, database poolsize, maximum number of

threads, etc.


26/32


Constant monitoring to check if the system is meeting demands: workload (peak periods, unusual patterns)

external metrics (user-perceived)

internal metrics (help to detect bottlenecks and to fine tune the system)

availability (external and internal)

May need to dynamically adjust configuration parameters


27/32


Systems may need to evolve to cope with new laws and Regulations(e.g., HIPPA)

Systems may need to evolve to provide new functions (e.g., sale of

downloadable MP3 music in addition to CDs)

How are the IT resources going to cope with evolution in terms ofSLAs?


28/32

Reference Model for IT

Business Model:- number of branches

- number and location of ATMs

- number of accounts of each type

-business evolution plans (e.g., mergers)

Social Model- privacy policy

- accessibility policy


29/32

User Model:

Customer Behavior Model Graph


30/32

IT Infrastructure: Example


31/32

Homework #1

Page 30, #1 and #2 try them, if you can

Page 30, #3 and #4 write them up


32/32

Tomorrow

Paper presentation session 1

scheduling, selection of papers, discussion of best

practices

Documents

Scheduling in CPU