Solid state disk (SSD) drive technology is rapidly gaining acceptance for high-end data storage applications because of the many benefits it offers enterprise users, including low power consumption and improved performance, compared to fibre channel and serial SCSI hard disk drives.
Although current SSD capacities are generally lower than conventional hard drive capacities, SSDs are attractive to users who are able to harness their performance advantage to address high input/output (I/O) demands.
SSDs are also appealing because they require less than half the power of a conventional hard disk drive. More significantly, they require one-tenth to one-one hundredth of the power per I/O operation - for both read and write transactions.
Against this backdrop, SSDs will most likely be used in storage environments where the data used in the application would benefit from very low latency access. Applications include databases that are required to run in memory for performance reasons, as well as business intelligence, data mining and virtual desktop infrastructure solutions.
Other targeted applications include content caching for near static data file/Web servers, real-time financial data processing and verification, seismic data processing, CAD/CAM, and 3D animation/rendering.
Make haste
Many industry watchers believe the speed with which organisations will adopt SSD technology into their data centres will eclipse the uptake of technologies linked to storage area networks (SANs) and serial ATA mass storage devices, including optical drives - despite the (fast falling) price premium currently associated with SSDs.
SSDs are also appealing because they require less than half the power of a conventional hard disk drive.
John Hope-Bailie is technical director of Demand Data.
Up to now, network and data centre managers faced with rapidly increasing I/O demands have tended to add capacity to hard disk storage arrays by simply adding more drives.
However, to reach the performance peaks demanded of today's applications, each drive must, of necessity, contain less data to accommodate the inherent mechanical latency of the drive and slow spindle speeds, with the result that capacity utilisation rates plummet significantly.
Traditional high-end hard disk drives' spindle speeds are around 10 000 to 15 000rpm, which do not allow them to keep pace with I/O requests in demanding applications such as online transaction processing, data warehouse queries and data analytics applications.
The future
Consequently, just a few SSDs, with their appreciably higher random read times, can replace rows of hard disk arrays, considerably reducing total cost of ownership. Replacing bulky hard disk arrays with small-footprint SSDs will also reduce hardware clutter in the data centre. A spin-off benefit will be reduced power consumption.
In a widely reported test undertaken by a leading hardware vendor, a 4.1TB SSD array, with a capacity of one million I/Os per second, was able to handle data warehouse queries five times faster than a competing array of 100 high-end 15 000rpm hard disk drives.
While SSD technology is seen by many as 'the future', there are factors, apart from the aforementioned price premium, which could impact its uptake in the marketplace. These include the varying quality of certain SSD devices on the market and the inability of certain brands/models to offer full end-to-end error correction or native encryption capabilities.
Obviously, purchasers of this technology need to research the market and put various suppliers' offerings to test before making a decision. Standards-based testing for specific environments can validate their expectations.
It is also worthy noting that the overall performance of any data storage solution will be impacted by internal bus speeds, read/write electronics, host port speeds and many other criteria.
Share