Customized Database And Systems Tuning Methodology: A Case Of Payroll Processing Optimization
View/ Open
Date
2013Author
Mukiri, Michael Njoroge
Type
ThesisLanguage
enMetadata
Show full item recordAbstract
Performance of database applications is becoming increasingly important as more and more processes are
automated in organizations. Most businesses rely heavily on database systems and non-optimal
performance of computer system has an immediate negative impact on business. With internet becoming
cheaper and millions of mobile phones being used to access systems, business applications are under
pressure to perform optimally and support business needs.
An evaluation and analysis of existing database and application tuning methodologies was conducted and
gaps identified. An integrated holistic database and application fine tuning methodology was developed
that tried to address the identified gaps. The aim of the methodology is to guide the tuning expert in
identifying bottlenecks in various tiers (Database, Network, and Application) that can have impact on
performance. In each of the tiers the main bottlenecks that should be looked out for are highlighted and
resolutions of how to resolve them are also suggested. The developed methodology makes no assumption
as to which tier contains the bottleneck but gives guidelines on how various bottlenecks can be identified in
each tier. In addition an overview of existing tools for gathering performance statistics and utilizations
statics for various tiers is given. To address the challenge of choosing which bottleneck to resolve first,
ranking formulae is suggested that ranks bottlenecks based on their overall expected improvement in
performance after resolution of the bottleneck, cost of resolving the bottleneck and the time taken to resolve
the bottleneck. The bottleneck with highest rank is resolved first.
The methodology was applied on a poorly performing payroll application and it was proven that response
time of the payroll application could be improved significantly by eliminating identified bottlenecks. By
use of the methodology it was illustrated how tuning experts can approach the tuning exercise in a more
structured and holistic approach with the sole purpose of identifying and resolving bottlenecks in any tier as
fast and cost effective as possible. The tuning developed database application tuning methodology can be
in-cooperated in database and systems administrators’ standard operating manuals as a tuning guide.
Publisher
School of Computing and Informatics