A new algorithm for computing joins with grid files

The BR/sup 2/-directory representation, a directory structure for grid files, and a join algorithm for the evaluation of general n-ary joins on grid files are presented. It is shown that the CPU cost of the join algorithm is successfully reduced by introducing an inner join. A comparison with the hash join algorithm and a join algorithm on k-d trees for equijoins is based on a cost model developed for query processing with grid files. The join algorithm outperforms the hash join, a specialized join method for equijoins, and the join algorithm on k-d trees for equijoins.<
>