Modelling Database Concurrency Control Algorithms using a general Purpose Performance Evaluation Tool
We present a performance evaluation study comparing two often-mentioned concurrency control algorithms, namely (r,x)-locks and OPTIMISTIC concurrency control as proposed by Kung and Robinson. This study was conducted by applying a general purpose performance evaluation tool, COPE. The results show that OPTIMISTIC concurrency control leads to worse database management system performance than LOCKING methods under a wide variety of workload conditions. The application of COPE pointed out how such a study in a very specialized field can be supported by a general tool, and what a general tool is missing for an application to specialized problems.
Nutzung und Vervielfältigung:
Alle Rechte vorbehalten