|
|
|
 |
|
 |
|
Exploiting Data Lineage for Parallel Optimization in Extensible DBMSs
|
E.C. Shek and
R.R. Muntz
View Paper (PDF)
Return to Poster Session 2: Interoperable Databases
Accommodation of complex user-defined data types and operations, and high query performance are among the most important requirements of advanced information systems. They together strongly suggest the need for extensibility and non-relational query support in parallel geoscientific query optimization, which have been recognized as open research problems. Parallelizing a query execution plan involves determining how input data streams to evaluators implementing logical operations can be divided to be processed by clones of the same evaluator in parallel. The fine-grained data lineage characteristics of user-defined evaluators dictates how they can be parallelized, and hence have to be systematically captured and made available to the parallel query optimizer. In this paper, we introduce a model for characterizing the data lineage properties of user-defined evaluators in an extensible DBMS, and describe a parallel query processing system that exploits such data lineage characteristics for parallel query optimization. We have demonstrated and validated our system with applications such as geoscientific data mining.
Copyright(C) 2000 ACM
|
|
|
|
|
|
|
