Maintaining large update batches by restructuring and grouping

Materialized views defined over distributed data sources can be utilized by many applications to ensure better access, reliable performance, and high availability. Technology for maintaining materialized views is thus critical for providing up-to-date results since a stale view extent may not help or even mislead these applications. State-of-the-art incremental view maintenance requires O(n²)$\mathrm{O}(n^2)$ or more remote maintenance queries with n being the number of data sources in the view definition. In this work, we propose two novel maintenance strategies, namely adjacent grouping and conditional grouping, that dramatically reduce the number of maintenance queries required to maintain the materialized views. This reduction in the number of maintenance queries brings the basic trade-off between the complexity of each query and the total number of maintenance queries that can be exploited to improve maintenance performance. The proposed maintenance strategies have been implemented in a working prototype system called TxnWrap. Experimental studies illustrate that our proposed strategies are able to achieve about 400% performance improvement in terms of total processing time compared with existing batch algorithms in a majority of cases.