NVMe queue management multi-tier storage systems

A system comprising:

• a processor;
• and
• a computer-readable storage medium communicatively coupled to the processor and storing program instructions which, when executed by the processor, are configured to cause the processor to:
• identify, by a storage system, a set of subsystems, wherein:
• the storage system:
• is a multi-tier storage system;
• stores a set of set of extents, wherein each extent is stored on one subsystem of the set of subsystems;
• includes a plurality of cores, wherein each core is assigned to process at least one extent of the set of extents, and each core of the plurality of cores is mappable to all subsystems of the set of subsystems;
• and
• each subsystem in the set of subsystems is:
• communicatively connected to the storage system via a non-volatile memory express (NVMe) protocol, and
• correlated to a tier of the multi-tier storage system;
• add a new subsystem to the storage system, wherein the adding includes establishing a connection to the new subsystem, retrieving, from the new subsystem, a set of capabilities of the new subsystem and assigning, based on the set of capabilities, the new subsystem to a tier of the storage system;
• generate a migration plan, in response to adding the new subsystem, wherein the migration plan is configured to move at least one extent of the set of extents between two or more subsystems of the set of subsystems including a first extent from a first subsystem of the set of subsystems to the new subsystem;
• determine a predicted workload of each extent in the set of extents;
• map, based on the predicted workload of each extent, each extent to a core of the plurality of cores, wherein the mapping is configured such that each core is balanced;
• and
• assign, based on the mapping, each extent to the mapped core, wherein the assigning includes establishing an input/output queue (IOQ) for each extent including a first IOQ for the first extent between the mapped core and the new subsystem.

A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processing unit to cause the processing unit to:

• identify, by a storage system, a set of subsystems, wherein:
• the storage system:
• is a multi-tier storage system;
• includes a plurality of cores, wherein each core of the plurality of cores are mappable to all subsystem of the set of subsystems;
• and
• each subsystem in the set of subsystems is:
• communicatively connected to the storage system via a non-volatile memory express (NVMe) protocol, and
• correlated to a tier of the multi-tier storage system;
• add a new subsystem to the storage system, wherein the adding includes establishing a connection to the new subsystem, retrieving, from the new subsystem, a set of capabilities of the new subsystem and assigning, based on the set of capabilities, the new subsystem to a tier of the storage system;
• receive, by the storage system, a tiering notification, wherein the tiering notification indicates at least one extent of a set of extents stored on the storage system will be migrated between subsystems;
• generate a migration plan, in response to the receipt of the tiering notification, wherein the migration plan is configured to move the at least one extent of the set of extents between two or more subsystems of the set of subsystems including at least a first extent from a first subsystem of the set of subsystems to a second subsystem of the set of subsystems;
• monitor a workload for each core of the plurality of cores;
• determine a predicted workload of each extent in a set of extents;
• rank, based on the predicted workload, each extent of the set of extents;
• map, based on the predicted workload of each extent, each extent to a core of the plurality of cores, wherein the mapping is configured such that each core is balanced, and each core of the plurality of cores are mappable to all subsystem of the set of subsystems, and where the highest ranked extent and the second ranked extent are mapped to different cores;
• establish, based on the mapping, an input/output queue (IOQ) for each extent including a first IOQ for a first extent on a first core of the plurality of cores and a second IOQ for a second extent on the first core, wherein each IOQ is processed by the mapped core of the plurality of cores;
• determine, in response to the monitoring, the first core of the plurality of cores is overloaded, and the first extent of the set of extents and the second extent of the set of extents are mapped to the first core of the plurality of cores;
• identify, in response to the monitoring, a second core of the plurality of cores, wherein a second workload of the second core is less than a first workload of the first core;
• and
• re-map, based on the identification, the second extent to the second core, wherein the re-mapping includes establishing a third IOQ for the second extent on the second core and terminating the second IOQ for the second extent on the first core.