An allocation of tables across a distributed database provides for taking into account existing table groups while also obtaining newly calculated table groups that are formed according to a new table group allocation. Such a new table group allocation may be desired to optimize workload balancing and/or data transfers for new workload patterns. Maintaining one or more types of existing table groups allows for the existing table groups to not be overwritten during allocation of the tables. An identification of tables is used, as well as a selected user-defined grouping option to indicate a mode for maintaining or replacing the existing table groups. An iterative process includes assigning, for each table, a group name and a group type.