#
Publication:

Speeding up Navigational Requests in a Parallel Object Database System (2002)

Author(s): Smith J, Watson P, de F Mendes Sampaio S, Paton NW

    Abstract: In data intensive applications, both programming and declarative query languages have attractions, the former in comprehensiveness and the latter for ease of use. Databases sometimes support the calling of side-effect free user defined functions from within declarative queries. As well as enabling more efficient coding of computationally intensive functions, this provision not only moves computation to data in a client-server setting, but also enables speedup through data parallel execution if the server is parallel. There has been little work on the combined use of query and program based database access in the context of parallel servers. We believe Polar is the first parallel object-oriented database which supports this arbitrary navigation both in a client application and in functions (operations) which may be called from within declarative queries. This work introduces Polar's support for navigation and for calling operations from within parallel queries and presents performance results for example navigational requests for the large OO7 benchmark.

      • Date: 27-30 August 2002
      • Conference Name: 8th International Euro-Par Conference (Euro-Par)
      • Volume: 2400
      • Pages: 332-341
      • Publisher: Springer
      • Publication type: Conference Proceedings (inc. abstract)
      • Bibliographic status: Published
      Staff

      Professor Paul Watson
      Professor of Computing Science