Improving OLAP Workload Performance in Apache Ignite

Abstract

Apache Ignite is an open-source system that is used to process database workloads. It utilizes Apache Calcite as the underlying query engine to parse SQL queries into relational algebra operators that are passed to the Ignite execution engine. This thesis investigates the performance of online analytical processing (OLAP) workloads with varying memory and data distribution settings using Apache Ignite with Apache Calcite. From empirical studies, practical strategies to decrease response times for OLAP queries are designed and implemented in Apache Ignite. Through studies using the TPC-H benchmark, it is demonstrated that each strategy yields performance improvements across multiple queries, and when all strategies are enabled, average performance improves generally for all queries in the workload.