Distributed query processing in fast local networks
A scheme to make use of semantic information to process distributed queries locally without data transfer with respect to the join clauses of the query is proposed. Since not all queries can be processed without data transfer, an algorithm to recognize the locally processable queries is given. For nonlocally processable queries, a simple fragment-and-replicate algorithm is used. The algorithm chooses a relation to remain fragmented at the sites where they are situated while replicating the other relations at those sites. The algorithm determines the chosen relation such that the minimum response time is obtained. If the fragments of the relations left fragmented are allowed to be processed on other sites, to choose the optimal processing sites is NP-hard. A heuristic is given for that situation. The problem of determining necessary local reductions (i.e., selections and projections) is shown to be NP-hard. A heuristic is provided. The integration of semi-joins with the fragment-and-replicate algorithm is also discussed. Some properties of profitable semi-joins are identified. When no relation referenced by a query is fragmented, an optimal algorithm is given to decide which relation and its copy is to be partitioned into fragments, how the relation is to be partitioned, and where the fragments are to be sent for processing. The conditions for efficient processing of aggregate operators are also discussed.
- Research Organization:
- Illinois Univ., Chicago (USA)
- OSTI ID:
- 7183518
- Country of Publication:
- United States
- Language:
- English
Similar Records
Distributed-query optimization in fragmented data-base systems
Knowledge-based approach to multiple-transaction processing and distributed data-base design