Providing efficient, extensible and adaptive intra-query parallelism for advanced applications

Parallel execution offers a solution to the problem of reducing the response time of object-rela-tional queries against large databases. A database management system answers a query by first finding a procedural plan to execute the query and subsequently executing the plan to produce the query result. In this thesis we address all significant levels of the query processing architecture in order to provide a comprehensive approach to the problem of efficient intra-query paral-lelism. Thereby, we develop optimization and parallelization algorithms using models that incorporate the sources of parallelism as well as obstacles to achieve speedup. To reduce its inherent complexity , we have split parallelization into several phases, each phase concentrating on particular aspects of parallel query execution. This rule-and cost-based approach guarantees both exten-sibility as well as effectiveness. Adaptability to diverse application domains and architectural characteristics are provided by means of appropriate parameter settings. The proposed strategies have been implemented and evaluated within the parallel object-rela-tional DBMS prototype MIDAS. The results show that the presented approach is particularly suitable for the parallelization of large and complex queries, as can be found in upcoming applications such as data warehouses, digital libraries or stream analysis.