Managing SSD wear rate in hybrid storage arrays
A technique for use in balancing flash drive wear in data storage systems is disclosed. Multiple sets of flash drives are identified where data is stored as multiple slices striped across the set of flash drives. A write rate at which data will be written to the multiple slices stored on the set of flash drives during a next time interval is predicted. A number of bytes that can be written to each set of flash drives is determined. A metric representative of a wear rate is determined for each set of flash drives. HDD relocation candidates are identified and a relocation process to relocate identified slices initiated.
1. A method for use in balancing solid state drives (SSD) wear in data storage systems, the method comprising:
identifying multiple sets of SSDs and multiple sets of hard disk drives (HDDs), wherein each set of SSDs and HDDs store data arranged in multiple slices striped across the respective set of SSDs and HDDs;
predicting, for each set of SSDs, a write rate at which data will be written to the multiple slices stored on the set of SSDs during a next time interval;
determining, for each set of SSDs, a number of bytes that can be written to each set of SSDs, wherein the number is based on a remaining program/erase (PE) cycle count for each respective set of SSDs;
determining, for each set of SSDs, a wear metric representative of a wear rate corresponding to the set of SSDs, the metric based on a SSDs' determined predicted write rate and the determined number of bytes that can be written to each set of SSDs;
identifying one or more HDDs as candidates for relocating slices stored on a set of SSDs, wherein the set of SSDs is predicted to wear out before reaching an end of an interval period; and
initiating a relocation process to relocate slices identified for relocation.
2. The method as claimed in claim 1 , wherein the end of an interval period is an end of warranty period associated with the SSD on which the slices identified for relocation reside.
3. The method as claimed in claim 1 , wherein SSD slices having a hot temperature are assigned a high priority for relocation.
4. The method as claimed in claim 1 , wherein SSD slices storing metadata are assigned a high priority for relocation.
5. A system for use in balancing SSD wear in data storage systems, the system comprising:
a storage processor and memory configured to:
identify multiple sets of SSDs, wherein each set of SSDs store data arranged in multiple slices striped across the set of SSDs;
predict, for each set of SSDs, a write rate at which data will be written to the multiple slices stored on the set of SSDs during a next time interval;
determine, for each set of SSDs, a number of bytes that can be written to each set of SSDs, wherein the number is based on a remaining program/erase (PE) cycle count for each respective set of SSDs;
determine, for each set of SSDs, a wear metric representative of a wear rate, the metric based on a respective SSDs determined predicted write rate and the determined number of bytes that can be written to each set of SSDs;
identify one or more HDDs as candidates for relocating slices stored on a set of SSDs, wherein the set of SSDs is predicted to wear out before reaching an end of an interval period; and
initiate a relocation process to relocate slices identified for relocation.
6. A computer-program product including a non-transitory computer-readable storage medium encoded with computer-program code that, when executed on a processor of a computer, causes the computer to manage data storage in a data storage system, the computer-program product further comprising:
identifying multiple sets of solid state drives (SSDs 1 and multiple sets of hard disk drives (HDDs), wherein each set of SSDs and HDDs store data arranged in multiple slices striped across the respective set of SSDs and HDDs;
predicting, for each set of SSDs, a write rate at which data will be written to the multiple slices stored on the set of SSDs during a next time interval;
determining, for each set of SSDs, a number of bytes that can be written to each set of SSDs, wherein the number is based on a remaining program/erase (PE) cycle count for each respective set of SSDs;
determining, for each set of SSDs, a wear metric representative of a wear rate corresponding to the set of SSDs, the metric based on a SSDs' determined predicted write rate and the determined number of bytes that can be written to each set of SSDs;
identifying one or more HDDs as candidates for relocating slices stored on a set of SSDs, wherein the set of SSDs is predicted to wear out before reaching an end of an interval period; and
initiating a relocation process to relocate slices identified for relocation.