Comparing IndexedHBase and Riak for Serving Truthy: Performance of Data Loading and Query Evaluation

摘要

This report summarizes our performance evaluation of IndexedHBase and Riak in their support of the Truthy system. We used data from June 2012 to test the performance of these platforms in two aspects: data loading and query evaluation. The total data size is 352GB. Queries tested can be found online. For IndexedHBase, we conducted scalability tests for data loading using more nodes on Alamo in FutureGrid (see appendix) while basic tests where done on 8 nodes of the FutureGrid Bravo cluster.