Automated Tiered Storage Gaining Traction Through Cost Savings, Performance


Automated tiered storage has become a nearly ubiquitous feature in the industry thanks to its perceived benefit of being able to move a company's data to different types of storage media based on the value of the data and the cost of the storage.

Most major storage vendors offer their own slightly-different take on automatic tiered storage either as an option or as an included feature, and they say uptake of the feature among customers is strong.

However, the feature is not without its detractors, with other storage capacity management solutions falling storage media prices providing alternatives to customers and their solution providers.

Automated tiered storage, or dynamic tiering, is the ability of a storage device to automatically migrate data from one type of media to another based on how "hot," or frequently accessed, the data is and on the cost of the different types of media.

With automated tiered storage, data which is frequently accessed by an application should be sitting on the fastest available media so that it can be read or updated quickly. As that data ages, or becomes less frequently accessed, it can be migrated to slower hard disk drives or eventually to tape or a storage cloud where the cost of storing it is much lower. However, that data can also be migrated back to higher-speed media if access to it increases.

The user should not be aware on which tier his or her data is sitting. However, the user may notice a drop in performance when accessing data which has not been accessed for some time.

Dynamic tiering is not a new technology. Originally known as hierarchical storage management, or HSM, it was first developed by IBM to migrate mainframe data between hard drive and tape.

Dynamic tiering takes advantage of the performance vs. cost characteristics of different types of storage media.

Tier 0, which is currently the fastest storage tier, includes solid state drives (SSDs). SSDs get their high performance from the use of high-speed Flash memory instead of spinning discs to store data. In an automated tiered storage environment, frequently accessed data may be kept on SSDs while all other data is moved to other tiers because of the high cost of SSDs.

Tier 1 includes high-speed hard drives, typically those with SAS or Fibre Channel interfaces. Tier 2 typically includes lower-speed SATA hard drives. SAS or Fibre Channel drives feature higher performance than SATA drives, but SATA drives are much less expensive on a per-gigabyte basis and feature much higher capacities.

Tier 3 storage includes low-cost tape and optical disc media. However, their large capacities and slow performance make them useful for backing up or archiving data which is not likely to be accessed for normal business purposes.

There is currently talk around an even higher performance tier of storage based on Flash or DRAM memory in servers which can be used as a cache in front of SSDs to increase the performance of accessing the hottest data.

EMC, for instance, in May unveiled its "Project Lightning," a new PCIe Flash-based server cache to move data closer to the processor. By extending EMC's Fully Automated Storage Tiering (FAST) technology to that cache, it will also serve as a high-performance direct-attach storage device. It is expected to be available later in 2011, and may be part of an automated tiered storage architecture after then.

NEXT: Different Vendors, Different Ways To Tier Storage