An ETL optimization framework using partitioning and parallelization

Extract-Transform-Load (ETL) handles large amounts of data and manages workload through dataflows. ETL dataflows are widely regarded as complex and expensive operations in terms of time and system resources. In order to minimize the time and the resources required by ETL dataflows, this paper presents an optimization framework using partitioning and parallelization. The framework first partitions an ETL dataflow into multiple execution trees according to the characteristics of ETL constructs, then within an execution tree pipelined parallelism and shared cache are used to optimize the partitioned dataflow. Furthermore, multi-threading is used in component-based optimization. The experimental results show that the proposed framework can achieve 4.7 times faster than the ordinary ETL dataflows (without using the proposed partitioning and optimization methods), and is comparable to the similar ETL tools.