A New Algorithm for Processing Joins Using the Multilevel Grid File

Join is an operation that is frequently used and the most expensive in processing database queries. In this paper we propose a new efficient join algorithm(called the MLGFJoin) for relations indexed by the tnul~ilevrl grirlfile(MLGF), a multidimensional dynamic hashed file organization. The MLGF-Join uses the domain space partition maintained in the directory of the MLGF. The MLGF-Join can process a join with one scan of the relations to be joined, assuming a main memory buffer is available that is sufficiently large for a range-oriented subjoin. Besides, the MLGF-Join does not require the costly preprocessing such as sorting in the sortmerge join algorithm and domain partitioning in the hash join algorithm. We also discuss the advantages of the MLGF-Join over others based on multidimensional dynamic file organizations such as the grid file, K-D tree, and multikey hashing.