Clustering relations of large databases for parallel querying
Bobbie · 1994
We discuss a technique for clustering relations for parallel querying in a distributed computing environment. Under the scheme, relations of relational database systems are grouped into clusters. Each cluster contains relations which are interrelated based on their primary-secondary key characteristics or commonality of their attributes/fields. Once grouped, the clusters are then distributed across a network of Sun SPARCstations, running the Parallel Virtual Machine (PVM) environment for parallel querying. The PVM system allows the total computing power of a collection of heterogeneous or homogeneous, computing processors to be harnessed for solving large problems. We have carried out parallel execution of SQL relational database queries using about 100 relations of a commercial application.>