US 11,886,732 B2
Data storage server with multi-memory migration
Paul F. Kusbel, Longmont, CO (US); and John E. Moon, Superior, CO (US)
Assigned to Seagate Technology LLC, Fremont, CA (US)
Filed by Seagate Technology LLC, Cupertino, CA (US)
Filed on Jan. 31, 2017, as Appl. No. 15/420,785.
Prior Publication US 2018/0217778 A1, Aug. 2, 2018
Int. Cl. G06F 3/06 (2006.01)
CPC G06F 3/0649 (2013.01) [G06F 3/061 (2013.01); G06F 3/0685 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
storing a data packet generated by a client in a first memory of a data storage server having the first memory, a second memory, a third memory, a controller, a data migration module, and a prediction module, each of the first, second and third memories having a different data transfer latency and associated cost of data storage, the data migration module configured to reactively identify a minimum data transfer latency required for the data packet to satisfy a quality of service (QoS) agreement between a host and the client based on logged performance metrics associated with the data packet, the prediction module configured to proactively predict a change in future access requirements for the data packet by the client based on predictive performance metrics associated with the data packet;
copying the data packet to the second memory proactively as directed by the prediction module;
concluding, with the data migration module, that storage of the data packet in the second memory will satisfy the QoS agreement;
determining, with the data migration module, that storage of the data packet in the third memory will satisfy the QoS agreement; and
migrating the data packet to the third memory with the controller as directed by the data migration module in response to the third memory providing storage of the data packet with a lower cost of data storage to the client than the second memory, and in response to the prediction module predicting no change in the future access requirements for the data packet by the client based on the predicted performance metrics.