Optimizing parallel query plans and execution

Author

Leslie, H.

Author_Institution

Tandem Comput. Inc., Cupertino, CA, USA

fYear

1991

fDate

Feb. 25 1991-March 1 1991

Firstpage

105

Lastpage

109

Abstract

It is noted that Tandem´s loosely coupled multiprocessor machine is a natural environment for parallel database operations. Horizontally partitioned tables are supported by the Tandem system. It is shown how this environment is utilized in Tandem´s NonStop SQL implementation. The Tandem architecture, parallel index maintenance, parallel use of partitioned tables, and repartitioning are discussed. It is shown that the use of the parallel execution in NonStop SQL can result in a large speedup in the execution of operations on partitioned and nonpartitioned data. A benchmark showed that parallel queries in NonStop SQL achieve linear speedup and scaleup for batch queries.<>

Keywords

database management systems; parallel architectures; query languages; Tandem architecture; Tandem system; Tandem´s NonStop SQL; batch queries; benchmark; horizontally partitioned tables; linear speedup; loosely coupled multiprocessor machine; parallel index maintenance; parallel query plan optimisation; repartitioning; scaleup; Aggregates; Availability; Computer architecture; Concurrent computing; Data flow computing; Databases; Electrostatic precipitators; Operating systems; Process control; Transfer functions;

fLanguage

English

Publisher

ieee

Conference_Titel

Compcon Spring '91. Digest of Papers

Conference_Location

San Francisco, CA, USA

Print_ISBN

0-8186-2134-6

Type

conf

DOI

10.1109/CMPCON.1991.128791

Filename

128791