Performance Analysis and Benchmarking of a Database with Cached Content

This report describes how a database benchmark metric can be defined and also important aspects of how to tune a database server. The starting point is a database which content is cached in main memory by a caching server. In some situations, bad write performance has been noticed and the aim is to tune the database within its current data model and also to define benchmarks metrics. Therefore the database and its surrounding system have been analyzed. This analysis and previous research in the benchmark field have lead to the definition of a benchmark metric. The database has then been optimized and performance gains have been documented with the developed benchmark. The results show that much can be done to increase performance. Thus, the performance gains are in several situations rather small. To achieve larger performance gains, the data model must probably be analyzed in depth and changed accordingly. Analys och prestandamätning av en databas med tillhörande cachingserver