PVM: Parallel view maintenance under concurrent data updates of distributed sources

Abstract

Data warehouses (DW) are built by gathering information from distributed information sources (ISs) and integrating it into one customized repository. In recent years, work has begun to address the problem of view maintenance of DWs under concurrent data updates of different ISs. The SWEEP solution is one solution that does not require the ISs to be quiescence, as required by previous strategies, by employing a local compensation strategy. SWEEP however processes all update messages in a sequential manner. To optimize upon this sequential processing, we now propose a parallel view maintenance algorithm, called PVM, that incorporates all benefits of previous maintenance approaches while offering improved performance due to parallelism. We have identified two issues critical for supporting parallel view maintenance: (1) detecting maintenance-concurrent data updates in a parallel mode, and (2) correcting the problem that the DW commit order may not correspond to the DW update processing order due to parallel maintenance handling. In this work, we provide solutions to both issues. We have implemented both SWEEP and PVM in our EVE data warehousing sys- tem, and our studies confirm the multi-fold performance improvement of PVM over SWEEP.