Interleaving a join sequence with semijoins in distributed query processing

The problem of combining join and semijoin reducers for distributed query processing is studied. An approach based on interleaving a join sequence with beneficial semijoins is proposed. A join sequence is mapped into a join sequence tree first. The join sequence tree provides an efficient way to identify for each semijoin its correlated semijoins as well as its reducible relations under the join sequence. In light of these properties, an algorithm for determining an effective sequence of join and semijoin reducers is developed. Examples are given to illustrate the results. They show the advantage of using a combination of joins and semijoins as reducers for distributed query processing