Parallel Hash Join, Parallel Merge Join, Parallel Nested Loops Join
Parallel join algorithms
These join algorithms are parallel versions of the traditional serial join algorithms. They are designed to exploit multiple processors on a network, within a machine, or even within a single chip.
Parallel join algorithms are based on the traditional serial join algorithms, namely (index) nested loops join, merge join, and (hybrid) hash join. The goal of parallel execution is to reduce the input sizes in each processing element, thus reducing the time for query completion even at the expense of increasing overall query execution effort due to data movement. Ideally, parallel join algorithms exhibit linear speed-up and linear scale-up.
Parallel join algorithms are orthogonal to pipelining among join operations in a complex query execution plan. Even a query with a single join can benefit from a parallel join algorithm. The essence of pipelining (and also of “bushy parallelism” in appropriate query execution plans) is to execute...