MPT: Suite Tools to Support Performance Tuning in NoSQL Systems

摘要

NoSQL databases are considered as a serious alternative for processing data whose volume reaches limits that are difficult to manage by relational DBMS. So far, they are praised for the capability to scale, replication and their capability to deal with new flexible data models. Most of these systems are compared to read/write throughput and their ability to scale. However, there is a need to get more in depth to monitor more precise metrics related to RAM, CPU and disk usage. In this paper, we propose a benchmark suite tools that enables data generation, monitoring and comparison. It supports several NoSQL systems including: column-oriented, document-oriented as well as multistores. We present some experimental results that show its utility.