Abstract: A computer compares an old database source file and new database source file to identify one or more changes between the old and the new database source files. Then, the computer generates one or more keys for each identified change. The keys are sorted into files that are then used to build a first dataset that includes at least one or more execution members. The execution members are initially grouped into one or more groups based on an order of execution of jobs associated with the one or more execution members. Then, within each group, the computer separates execution members that executable in parallel. Further, the computer builds a second dataset that includes at least a submit member for each group of execution members that executable in parallel. The submit member comprises a set of instructions to submit, for execution, all execution members of a group in parallel.