Performance of Semantic Caching Revisited
María Arinbjarnar, Bjarnsteinn Þórsson, Björn Þór Jónsson · Skemman · 2006
A caching architecture for database clients called semantic caching was proposed in 1996 and evaluated against a then-current relational database server 1998. While semantic caching was shown to perform well for a range of workloads, the relational server was not well equipped to handle complex remainder queries. Since then, hardware has become increasingly faster with considerable increases in memory size and caching capabilities. Additionally, there have also been significant performance improvements in relational database systems, in particular query optimization. In this report we first review some recent related work. We then evaluate the semantic caching architecture against modern hardware and software, and propose and evaluate two new approaches to query execution at the relational server. Our conclusion is that despite the hardware and software performance improvements which have reduced query processing time vary significantly, complex query workloads still present significant difficulties for the semantic caching architecture.