Adaptive Request Management Systems

RNDr. Miroslav Křipač, Ph.D.

Doctoral thesis

Abstract:

Web-based technologies are currently able to improve quality of services provided by information systems. System architects can benefit from easier management of applications and users appreciate wider data access. However, building information infrastructure strictly on web browsers may bring additional problems. One of these complications, which occur mainly inside large-scale information systems with strictly on-line transaction processing is that systems are opened potentially to a high number of users who access the system concurrently, which may easily cause system overloading. The first goal of this thesis is to propose alternative desirable architecture of large-scaled web-based information system, which is able to scale each resource easily according to its actual needs. Still, some situations, in which system load abnormally grows, cannot be resolved by adding more computational resources due to huge waste of investments. For this reason, the main part of this work focuses on alternative methods of automatic system adaptation to actual user behaviour. In this approach, a system can detect abnormal request load on-line and automatically adapt its request management to prevent overloading and ensure sufficient response to relevant requests. This thesis also includes possible methods of automatic system state detection, request classification and request management techniques.

Keywords

Web-Based Information Systems, Overloading, Adaptive Request Management, Mutli-Tiered Database Clusters

Language used: English

Date on which the thesis was submitted / produced: 9. 2. 2007

Thesis defence

• Date of defence: 9. 2. 2007
• Supervisor: prof. RNDr. Jiří Zlatuška, CSc.

Masaryk University

Faculty of Informatics

Doctoral programme / field:
Informatics / Informatics