For very big datasets, it is sometimes worthwhile to partition the datasets into smaller pieces.
However, partitioning leads to suboptimal results, even if the pieces are solved optimally:
A future version of OptaPlanner will have native support for several forms of partitioning. Meanwhile, you can
implement it yourself as shown in the image above. Use an OptaPlanner Solver to solve each
piece.
Note
Not all use cases can be partitioned. It only works on use cases for which the planning entities and value ranges can be divided into n pieces, such that none of the constraints crosses piece boundaries.