Skip to content

Latest commit

 

History

History
34 lines (29 loc) · 1.83 KB

Performance comparison between five NoSQL databases.md

File metadata and controls

34 lines (29 loc) · 1.83 KB
Raw

Performance comparison between five NoSQL databases

Citation

Tang, E. and Fan, Y., 2016, November. Performance comparison between five NoSQL databases. In 2016 7th International Conference on Cloud Computing and Big Data (CCBD) (pp. 105-109). IEEE.

Contributions

Compares the performance of 5 NoSQL databases (Redis, MongoDB, Couchbase, Cassandra, and HBase) through 2 experiments on a 4-node cluster.

  • Result
    • Experiment 1: Compares execution speed of the databases under certain conditions.
      • Data loading
        • Redis got the best performance.
        • Column-family databases, Cassandra and HBase, were slightly lower than Redis.
        • Couchbase presented the worst performance.
      • Workloads execution
        • Redis also got the best performance.
        • Column-family databases, Cassandra and HBase, were the slowest.
        • Document databases, MongoDB and Couchbase, presented good performance.
    • Experiment 2: Compares storage expansion tolerance of the databases.
      • Data loading
        • Redis was increasing at the fastest pace.
        • The throughput of Cassandra and Couchbase grew more rapidly compared to HBase and MongoDB.
      • Workloads execution
        • Redis still showed the best throughput performance.
        • Couchbase, Cassandra and HBase presented a similar trend as the loading stage (insertion process).
        • Throughput and efficiency of MongoDB decreased when the record exceeded a certain count limit.
  • Conclusions
    • Redis is particularly well suited for loading and executing workloads but has its limitation when facing extremely large data.
    • Document databases, followed by column-family databases, have a good average performance.
    • Database framework in master-master mode, where each node is equivalent, has more advantages over those in master-slave architectures.