View : 20 Download: 0

A STUDY ON THE BATCH AND MULTIPROGRAMMING SYSTEMS EVALUATION THROUGH SIMULATION

Title
A STUDY ON THE BATCH AND MULTIPROGRAMMING SYSTEMS EVALUATION THROUGH SIMULATION
Authors
李淑子
Issue Date
1975
Department/Major
대학원 수학과
Publisher
이화여자대학교 대학원
Degree
Master
Advisors
朴贊謨
Abstract
As computers have grown both in their sizes and specds, and their uses have become more versatile, computer system evaluation has become more important. Simulation technique has been widely used as one of the approaches in such evaluations. Mathematical models of the Batch Processing System and Multiprogramming System are constructed for computer system evaluation, and FORTRAN simulators for those models are developed. The system considered consists of a card reader (CR), a central processing unit (CPU), a line printer (LP), a disk, a tape, and a central memory (CM). The simulators are composed of four routines; they are initialization routine, scheduling routine, event routine, and queue update routine. A job moving through the system is marked by the occurrence of a series of events. In the Batch Processing System there exist 9 event routines, while there exist 15 event routines in the Multiprogramming System. For simulation of jobs processed by the system, two types of parameters are required. One type is system parameters such as speed of the card reader, speed of the line printer, and main memory size and the other job parameters such as job number, job type, job interarrival time, number of input cards, number of output lines, number of disk requests, number of tape requests, card reader processing time, CPU processing time, line printer processing time, disk processing time, tape processing time, job priority, interrequest time, arrival time, finish time, overhead time, system time, and memory space. The system parameters are read in as input data while the job parameters are generated when the job arrives to the system. Three objective functions, such as throughput time, turnaround time, and system utilization are employed for measurements of computer system performance. In the Batch Processing System, five test runs are made with selected sets of parameters. The evaluation is accomplished by analyzing the output results from the simulation runs; the simulator gives measurement of objective functions, job statistics, time statistics, facility utilization, queue statistics and frequency distribution of transit times. Six experimental runs are made with the Multiprogramming System each with different sets of parameters. Besides those outputs described in the Batch Processing System statistics on memory utilization is also obtained. From the results of the experimental runs with the given sets of parameters it is found that the computer system performance is affected by the speeds of the card reader and line printer and memory size considerably and by the arrival rate of the jobs, the job mixes, and assignment of priorities moderately. With the same sets of parameters the Multiprogramming system performed much efficiently than the Batch Processing System as expected. ;Computer가 속도와 크기면에서 발달하고, 그 용도가 다양해 짐에 따라 Computer System 평가가 더욱 중요하게 되었다. 평가 방법 중에서 가장 널리 사용되는 것의 하나가 Simulation기법이다. Computer System을 평가하기 위해 Batch Processing System과 Multiprogramming System의 수학적 모델을 설정했고, 각 system마다 simulator를 개발하였다. 이 system들은 card reader (CR), central processing unit (CPU), line printer (LP), disk, tape, 그리고 central memory (CM)로 구성되었으며 simulator 는 4개의 routine 즉 initialization routine, scheduling routine, event routine, queue update routine으로 구성되었다. 이 system을 통과하는 job의 상태는 일련의 event의 발생으로 나타내어진다. Batch Processing System은 9개의 event routine을 가지며, Multiprogramming System 에는 15개의 event routine이 있다. 이들 system에 의하여 처리되는 job을 simulate하기 위하여는 2가지의 parameter가 필요하다. 그 하나는 System parameter는 card reader속도, line printer 속도, main menory size 로서 input data로 읽혀지며, job parameter 는 job number, job type, interarrival time, number of input cards, number of output lines, number of disk requests, number of tape requests, card reacer processing time, CPU processing time, line printer processing time, disk processing time, tape processing time, job priority, interrequest time, arrival time, finish time, overhead time, system time, memory space로서 job이 이 System에 도착할 때 발생 되도록 하였다. system performance의 측정으로는 3개의 objective function을 사용했다. 즉 throughput time, turnaround time, 그리고 system utilization이 그들이다. Batch Processing System에서는 선택된 parameter set을 써서 5번의 test run을 수행하였다. 평가는 각 simulation run에서 얻어진 output results적 objective function 의 측정치, job statistics, facility 사용율, queue statistics, time statistics등을 분석하므로서 이루어 진다. Multiprogramming System 에서는 6번의 experimental run이 수행되었다. 위의 Batch Processing System에서 얻은 결과 외에 memory utilization에 대한 통계를 첨가하여 얻었다. 선정된 parameter를 가지고 시행한 experimental run에서 얻어진 결과를 통해볼때 card reader와 line printer의 속도, 및 memory size가 system performance에 많은 영향을 끼쳤으며 job의 arrival rate, job mix, 및 priority의 설정등은 그리 큰 영향을 끼치지 않았다. 같은 set의 parameter를 사용했을때 기대했던 대로 Multiprogramming System의 performance가 Batch Processing System보다 훨씬 좋았다.
Fulltext
Show the fulltext
Appears in Collections:
일반대학원 > 수학과 > Theses_Master
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE