When to use flash SSD instead of SAS or SATA?

Criteria for deploying flash solid state drives (SSD) are no different to any disk technology. Decision-making should match disk performance to application requirements based on three primary factors: cost, performance and capacity.

• The cost of flash SSD is much higher than that of SAS (approximately 10x), which is, in turn, greater than SATA.
• The sustainable I/O performance of flash can be 40 to 60 times greater than that of SAS and SATA.
• The standard capacity of SSD has increased to 600GB, but remains much lower in comparison to SAS (900GB) and SATA (3TB).

SSD is therefore of interest when an application's performance requirements outweigh those of cost and capacity, and is especially well suited to applications -- often the most business critical -- with the highest random read and write requirements.  

In some cases organisations have dramatically reduced datacentre footprint and storage costs, not to mention power and cooling needs, by replacing hundreds of SAS drives with a much smaller number of strategically-placed flash SSDs.

Flash drives should, therefore, be an important consideration for any organisation's tiered storage strategy, although in order to extract the greatest value from them it is important they are deployed in conjunction with the array maker's thin provisioning and automated storage tiering products.

Thin provisioning can greatly improve the utilisation of SSD capacity and avoid costly 'white space' as only actual data is written to the drives.  

Automated storage tiering promotes data that requires the greatest performance to the flash SSD tier, and moves less demanding data to SAS or SATA.

This use of different drive technologies together with thin provisioning and automated tiering maximises the effective use and capacity of the SSDs without the need for operator intervention.

David Boyle is senior consultant at GlassHouse Technologies (UK)