Skip to Main content Skip to Navigation
Theses

Étude des performances de méthodes de groupement dynamiques dans les bases de données orientées objet

Abstract : This work lies in the field of Object-Oriented Database Management System (OODBMS) performance evaluation. These systems fail to take over previous DBMSs (mainly relational systems, which are used since the 70's and well optimized) because of performance issues.

Several methods are used to enhance performances, such as object clustering. However, these techniques induce some overhead for the system. It is this primordial to evaluate their actual impact on global performance.

We have thoroughfully studied this domain to identify the major deficiencies in existing performance evaluation techniques (mathematical analysis, benchmarking, simulation). Our proposals, the Object Clustering Benchmark (OCB) and the Virtual Object-Oriented Database (VOODB) simulation model, aim at addressing these issues as generic, tuneable tools that are adapted to studying object clustering.

The VOODB simulation model integrates the OCB benchmark. It has been designed by applying a modeling methodology that aims at generalizing the analysis of OODBMSs and providing reliable models for these systems. Its code has been specified within a simple and powerful simulation environment, DESP-C++, we designed for this sake.

A validation process for our tools has been driven by measuring the actual performances of the O2 OODBMS, the Texas persistent object store and the DSTC dynamic object clustering technique, with OCB. Simulation performance evaluations of these systems have also been conducted with VOODB. Real-life and simulated results have been compared and proven consistent.
Document type :
Theses
Complete list of metadata

Cited literature [32 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00143351
Contributor : Jérôme Darmont <>
Submitted on : Wednesday, April 25, 2007 - 12:41:15 PM
Last modification on : Wednesday, April 21, 2021 - 8:52:04 AM
Long-term archiving on: : Wednesday, April 7, 2010 - 12:06:59 AM

Identifiers

  • HAL Id : tel-00143351, version 1

Citation

Jérôme Darmont. Étude des performances de méthodes de groupement dynamiques dans les bases de données orientées objet. Interface homme-machine [cs.HC]. Université Blaise Pascal - Clermont-Ferrand II, 1999. Français. ⟨tel-00143351⟩

Share

Metrics

Record views

458

Files downloads

641