In OODB, path expressions are usually used for navigating through complex objects and obtaining necessary information at the far end of a sequence of object references. The parallel forward pointer chasing algorithm (PFPC) and parallel backward pointer algorithm (PBPC) are natural algorithms for path expressions. If we change the implicit joins in the path expression into explicit joins, parallel join algorithms can also be employed for path expressions. In order to improve performance in a distributed environment, semi-join operations replace the join operations, which is called the parallel cascade semi-join algorithm (PCSJ). This paper mainly describes and compares PFPC and PCSJ algorithms implemented on DSVM; then analyzes the performance of algorithms from test results according to a certain benchmark. We can see PCSJ has much better performance than PFPC by the results. These parallel algorithms exploit pipelining and I/O communication parallelism; making full use of DSVM characteristics. Some valuable ideas are obtained from the theory analysis and practical testing.
[1]
Masaru Kitsuregawa,et al.
Parallel GRACE hash join on shared-everything multiprocessor: implementation and performance evaluation on symmetry S81
,
1992,
[1992] Eighth International Conference on Data Engineering.
[2]
David J. DeWitt,et al.
Parallel pointer-based join techniques for object-oriented databases
,
1993,
[1993] Proceedings of the Second International Conference on Parallel and Distributed Information Systems.
[3]
Tetsuro Kakeshita,et al.
Research on High Performance Databases
,
1999
.
[4]
David J. DeWitt,et al.
Parallel database systems: the future of high performance database systems
,
1992,
CACM.
[5]
Donovan A. Schneider,et al.
The Gamma Database Machine Project
,
1990,
IEEE Trans. Knowl. Data Eng..