Mr. Phanuel Seli Kwadwo Asense

Senior Systems Analyst


Research Areas/Interests

My research areas include developing algorithms to enhance processes in pieces of device like routers, switches etc. Also enjoys modelling business p...~more

My full CV

An Algorithm For Improving the Task Time Estimation in Software Application Installation

ABSTRACT:

It is a common experience how undesirable and frustrating it is, to wait minutes or hours unending for a software installation process, originally estimated to take a short time to complete. It is even more unacceptable to have a notification such as “0 seconds more” and the installation processes keep running into minutes. The objective of this article is to outline the shortcomings of the existing ways of estimating software installation times in order to suggest an alternative to improving the estimation process or algorithm.

 

Keywords : algorithm, installation, estimation, process

 

INTRODUCTION

Works get done in our world through the execution of processes. Processes have invariably become a part of our lives. It is often very unpleasant to a computer user to wait minutes or hours unending for a software installation process, originally estimated to take a short time to complete. It is even more frustrating to have a notification such as “ 0 seconds more” and the installation processes keep running into minutes. These issues definitely point out vividly that, the total estimated time users usually see have a wide margin of errors often because estimations usually take into consideration stand-alone time, Barsness et al. , (1995) and are done holistically. Furthermore, as the installation continues, the continual remnant time estimations are also with wide error margins. There are a lot of dynamic and non-dynamic factors that play out when running computer processes, ( Barsness et al. , 1995; Rao et al . , 1995).The approach of trying to estimate the total software installation process time holistically or globally and also considering only stand-alone time has a tendency of deviating from the actual installation time with a bigger error margin. Rather, phasing/decomposing or modularizing the time estimates gives a better result, MacGregor and Armstrong (1994). Also using the historical behaviour of the machine with respect to time, gives better time estimates, Barsness et al. , (1995). This article seeks to come up with a model or an algorithm that to a greater extent, is expected to minimize these margin of errors.

 

 THE CONCEPT OR ALGORITHM

In the opening paragraphs, it was pointed out that, trying to give time estimates for an installation process holistically has a tendency of deviating from the actual installation time with a bigger error margin than phasing or modularizing the time estimates. In view of this, the study adopted the approach of estimating the time in phases i.e. the ‘decomposition approach’ and also by factoring into the subsequent estimation processes, happenings from previous phases. This will allow the estimation algorithm to steady the machine’s behaviour, capability, environment and specifications. The algorithm will identify entities such as; the Installation device, Number and type of disk unit I/O processors, Processing unit (i.e. number and capacity), Disk unit space available, Main storage available in base pool and Licensed programs to be installed.

 

The user initiates the installation process by specifying user-defined settings or default / recommended settings. If this is done, a message is displayed indicating the installation task’s time estimation is underway. This message is given parallel to the installation process. Key domains/phases for the installation, based on user-defined or default settings are defined. Time for the first phase is estimated using the existing approach. The existing approach being the stand-alone time. At this point, the estimated time could have a wide margin of error but this would be rectified as the installation progresses. As the estimated time for phase one is displayed, the algorithm will produce alongside a notification assuring the user of the time for the remnant tasks. The times taken for distinct processes as the installation runs in phase one is logged. Times of historic situations that are similar to processes in succeeding phases are used in subsequent estimations to render better time values as installation progresses.

 

 

DISCUSSION

The fundamental concept based on which the earlier algorithm for estimating software installation time was built in no doubt had visible wide margin of errors. The new algorithm from the study even though had not been implemented yet, does show more efficient and potent ways of estimating than the latter.

 

CONCLUSION

 

To achieve more precision in estimating the time for software installation process is a daunting task owing to the array of complexities that spring up as regards the machine’s specifications, actual files to be processed, interrupts and what have you. In spite of this, there is also the need to better estimate the installation time since user frustration is also undesirable. An overview of the existing approaches pointed out factually that, most of the algorithms estimate the whole installation process time holistically, and in most cases do not factor similar historic transactions of the machine in the estimation process. The study’s approach uses the “divide and rule” tactics. This approach also gives chance to the algorithm to learn from earlier similar transactions to render better judgment or estimation as the installation progresses.

 

 

ACKNOWLEDGMENT

 

I am thankful to Dr. Osei Adjei of the Computer Science Department, Garden City University College, Kenyase and Dr. Evans Adei of the Department of Chemistry, KNUST for rendering me constructive criticisms that brought me this far. I am also grateful to the College of Science for allowing me use some of their facilities for the study.

 

 REFERENCES

 Barsness et al. , (1995) ‘Method And Apparatus For Estimating Installation Time In A Data Processing System’ pp.1-14

 J.A. Bergstra (ed.), A. Ponse (e.d), S.A. Smolka (e.d) , 2001, Handbook of Process Algebra , Publication Date: March 30, 2001 | ISBN-10: 0444828303 | ISBN-13: 978-0444828309 | Edition: 1

Page 5

 

Nathaniel Max Rock, 2005, Standards-Driven Math Vocabulary Ranking

Gulden Akinci, 2006, Private Tutor for Sat Math Success

 

 

Rao et al (1995) ’Method And Apparatus For Time Estimation And Progress Feedback On Distal Access Operations’   pp.1-14

 

Terry Connolly and Doug Dean (n.d.) ‘Decomposed Versus Holistic Estimates of Effort Required for Software Writing Tasks’ ; Department of Management and Policy, University of Arizona, Tucson, Arizona 85721 Center for the Management of Information ,College of Business and Public

Administration, University of Arizona, Tucson, Arizona8 5721

 

‘What is a process and processed-based management system?’ Available online at: www.netcoach.eu.com/index.php?id=38 [accessed 11th April, 2013]

 

‘What is a Process?’ Available online at: http://its.syr.edu/eps/services/process/what_is.html

[accessed 12th May, 2013]

 

 

 

 

 

 

 

 

 

 

 

Recent Blogs


Print page

SEND QUICK MAIL TO STAFF

Subject

Message

Your Email Address:
Note: this is to help the staff contact you back.