Performance Comparison Test of HBase and Cassandra Based on YCSB

摘要

As big data applications are becoming more and more popular, in view of the difficulties we had encountered in the application of traditional relational databases in the field of big data application, the column storage databases of non-relational databases are widely used in big data analysis system because of its high loading speed, real-time data loading, high compression rate and very suitable for aggregation operation, therefore, the read and write performance of column storage databases had became a hot topic in the field of big data application. In this paper, we compare the mainstream column storage databases(HBase and Cassandra), including analyzing the read and write data operation principles of the two databases, obtaining the performance results with the Yahoo! Cloud Serving Benchmark and analyzed the performance results. In this paper, we also discuss how to select the appropriate database and workload strategy to meet the requirements of system performance based on business needs. Experimental results show that Cassandra is superior to HBase in terms of single read/write performance. Cassandra is suitable for scenarios with high performance requirements for write data or single read/write data scenario, HBase had better read performance in mixed read and write data operation scenarios.