A Memory Allocation Model for Cassandra Database

Cassandra database system is one of the universal databases. To achieve high performance, we should allocate memory space rationally according to actual demands. To solve this problem, we firstly analyze the reading and writing process of Cassandra database and find out the corresponding memory space which will influence system performance. Secondly, we build up a relationship model between system performance and memory allocation. We have already applied the relationship model to real database servers for memory allocation and performance evaluation. Simulation results show that this memory model could describe the quantization relationship between memory space and system performance very well. It can help us get a higher performance for the Cassandra application.