An Asynchronous View Maintenance Approach Based on IO Sharing

Abstract

The materialized view is considered as a panacea that significantly facilitates query by trading space cost for execution time. However, it also involves the expensive cost of maintenance which trades away the latency. Disk IO cost is an important factor restricting view maintenance performance. To solve this problem, we decouple a complete procedure of view maintenance into a number of tasks and execute them asynchronously. We put off the executions of those tasks until necessary data access from disk can be completed by subsequent transactions or queries. The execution of view maintenance tasks and transactions or queries can share disk IO, which fundamentally reduces IO resource consumption and view maintenance overhead. We accomplish the complete architecture of view maintenance to support IO sharing. To maximize the advantages of sharing IO, the job of view maintenance is split into fine-grained tasks. An effective cost evaluator is also proposed to select proper tasks to be executed for subsequent transactions or queries. We implement our method on the PostgreSQL. Experiments on the CH-benCHmark show it achieves better performance than existing methods.