Performance and Replica Consistency Simulation for Quorum-Based NoSQL System Cassandra.

Distributed NoSQL systems such as Cassandra are popular nowadays. However, it is complicated and tedious to configure these systems to achieve their maximum performance for a given environment. This paper focuses on the application of a Coloured Petri Net-based simulation method on a quorum-based system, Cassandra. By analyzing the read and write process of Cassandra, we propose a CPN model, which can be used for performance analysis, optimization, and replica consistency detection. To help users understanding the NoSQL well, a CPN-based simulator called QuoVis is developed. Using QuoVis, users can visualize the read and write process of Cassandra, try different hardware parameters for performance simulation, optimizing system parameters such as timeout and data partitioning strategy, and detecting replica consistency. Experiments show our model fits the real Cassandra cluster well.