Optimized database partitioning

A method for optimized database partitioning, the method comprising:

• a database-management system (DBMS) comprising a processor, a memory coupled to the processor, and a computer-readable hardware storage device coupled to the processor, the storage device containing program code configured to be run by the processor via the memory to implement the method for optimized database partitioning, the method comprising:
• the system receiving a current workload of the DBMS, an execution log of the DBMS, and a set of data statistics,
• where the current workload comprises pending requests for access to records stored in a database managed by the DBMS,
• where the execution log lists results of servicing previous requests for access to records stored in the database,
• where the data statistics enumerate data values stored in records of the database;
• the system identifying a set of conditions,
• where each condition of the set of conditions is a logical predicate that specifies a contiguous set of key values capable of being stored in one column of the database, and
• where the set of key values is a subset of a set of key values specified by a logical predicate of at least one access request of the current workload;
• the system deriving a set of key-value ranges by correlating contents of the current workload and the execution log, where each range of the set of key-value ranges specifies two distinct conditions, of the set of conditions, specified by a same access request of the workload;
• the system distributing the set of key-value ranges among a set of logical partitions;
• the system building a logical-to-physical partition map as a function of a relative number of the access requests that request access to each range of the set of key-value ranges,
• where each physical partition of a set of physical partitions is associated with a range of numbers of access requests, and
• where the partition map maps ranges distributed to a first logical partition, of the set of logical partitions, to a first physical partition, of the set of physical partitions, if a number of times that access requests of the current workload request access to the ranges of the first logical partition falls within the range of numbers associated with the first physical partition;
• and
• the system physically storing, in a particular physical partition, of the set of physical partitions, database records identified by the key-value ranges of one or more logical partitions of the set of logical partitions,
• where the one or more logical partitions have been mapped to the particular physical partition.

A computer program product, comprising a computer-readable hardware storage device having a computer-readable program code stored therein, the program code configured to be executed by a database-management system (DBMS) comprising a processor, a memory coupled to the processor, and a computer-readable hardware storage device coupled to the processor, the storage device containing program code configured to be run by the processor via the memory to implement a method for optimized database partitioning, the method comprising:

• the system receiving a current workload of the DBMS, an execution log of the DBMS, and a set of data statistics,
• where the current workload comprises pending requests for access to records stored in a database managed by the DBMS,
• where the execution log lists results of servicing previous requests for access to records stored in the database,
• where the data statistics enumerate data values stored in records of the database;
• the system identifying a set of conditions,
• where each condition of the set of conditions is a logical predicate that specifies a contiguous set of key values capable of being stored in one column of the database, and
• where the set of key values is a subset of a set of key values specified by a logical predicate of at least one access request of the current workload;
• the system deriving a set of key-value ranges by correlating contents of the current workload and the execution log, where each range of the set of key-value ranges specifies two distinct conditions, of the set of conditions, specified by a same access request of the workload;
• the system distributing the set of key-value ranges among a set of logical partitions;
• the system building a logical-to-physical partition map as a function of a relative number of the access requests that request access to each range of the set of key-value ranges,
• where each physical partition of a set of physical partitions is associated with a range of numbers of access requests, and
• where the partition map maps ranges distributed to a first logical partition, of the set of logical partitions, to a first physical partition, of the set of physical partitions, if a number of times that access requests of the current workload request access to the ranges of the first logical partition falls within the range of numbers associated with the first physical partition;
• the system physically storing, in a particular physical partition, of the set of physical partitions, database records identified by the key-value ranges of one or more logical partitions of the set of logical partitions,
• where the one or more logical partitions have been mapped to the particular physical partition;
• and
• the system directing a storage manager component of the DBMS to split or combine partitions, of the set of physical partitions, as required to ensure that all physical partitions, of the set of physical partitions, are of a size that falls between a predetermined maximum size and a predetermined minimum size.