Multi-master architectures for distributed databases

A system for configuring a server set to process a workload of a data set according to performance criterion, the system comprising:

• one or more processors;
• memory in electronic communication with the one or more processors;
• instructions stored in the memory, the instructions being executable by the one or more processors to:
• for a first subset of data items of the data set, partition the server set into:
• a first master subset of at least two servers as masters that are permitted to update a data item from the first subset of data items, and
• a first non-master subset of remaining servers of the server set as non-masters that are not permitted to update the data item, wherein the designation of the first master subset and the non-master subset of the server set is based on fulfilling the performance criterion of the data set, and wherein at least one server from the first master subset is designated as a non-master with respect to a second data item from a different subset of data items of the data set than the first subset of data items;
• configure servers of the non-master subset to:
• receive a request to update the data item, and
• forward the request to another server that has been designated as a master of the data item;
• and
• configure the at least two masters of the first master subset to:
• receive the request to update the data item;
• update the data item according to the request;
• and
• propagate the update of the data item to at least one other server of the server set.