Jan. 23, 2008 | Tiered storage has been at the heart of data management strategies for many years. Now, it is increasingly important as new experiments and lab equipment produce larger volumes of data, and more of that data needs to be kept online and available (as opposed to archiving it offline on tape).
With tiered storage, the challenge has always been where to physically put data so that it is on a device with suitable performance characteristics to match an application’s needs. Naturally, a cost/performance analysis is involved: Simply putting all data on the highest-performing systems would be quite pricey.
Complicating matters is a new cost factor: energy efficiency. As is the case with servers, newer, higher-performance storage devices and systems typically use more electricity. These systems often incorporate the newest processors and disk arrays, both of which normally consume more energy.
This means a tier 1 storage system not only costs more per byte of storage space; it very likely also costs more for the electricity to operate it. This additional cost would have to factor into a data migration strategy, thus making it even more important to move data from tier 1 to tier 2 or lower storage systems whenever possible.
To put the energy consumption into perspective, some industry analysts estimate that each tier level of storage consumes about 1/8th the power of the next highest tier. So a tier 2 system, on average, would consume 1/8th the power of a tier 1 system and a tier 3 system (possibly used for archiving) would consume about 1/8th the power of a tier 2. (This means a tier 3 system would use about 1/64th the power of a tier 1 system.)
To address these electrical cost issues, storage system vendors are taking steps to cut energy consumption. Many are turning to more energy-efficient CPUs and power conservation techniques like throttling down the spin rate of unused disk drives.
Last week, EMC took another step to help reduce power consumption of its storage systems. The company announced it would integrate solid state drives (SSDs) into its core product line starting with the EMC Symmetrix DMX-4 storage system.
The SSDs drives have no mechanical parts and use flash memory to store and retrieve data. This not only makes them much faster than hard disks drives, but they also require much less power to run. EMC claims such flash drives can store a terabyte of data using 38 percent less energy than traditional mechanical disk drives. Still, flash drives have traditionally cost significantly more than disk storage, so the cost/performance analysis remains tricky.
What do you think of the use of flash memory and SSDs in storage systems? Do you routinely take storage system energy consumption into account when managing your data? Are you concerned about the power issues related to your storage systems? Drop me a line at s.salamone@att.net and share your thoughts on the subject.
------------------------
This article first appeared in Bio-IT World’s Inside IT newsletter. Click here for a free subscription to Inside IT.