Tiered storage the next big thing?

In a January 2014 'Ovum View' titled SAP refines Big Data strategy, author Tony Baer dissects SAP's HANA platform and the vendor's big data strategy, citing an Ovum research note titled “Storage tiering is the new black for databases” to share that the Ovum prediction for analytic applications is "aggressive use of data-tiering strategies, where data is stored on the appropriate medium (e.g., DRAM, SSD Flash, disk, tape) based on its utilization, or 'temperature.'"

The concept of storage tiering is simple. The idea is that the data available to an application can be separated or tiered into categories depending on how quickly it is required. So data that is required quickly should be stored somewhere you can get it quickly, while data that may not be required for years can be stored somewhere which may take time to retrieve. 

A simple analogy could be that the shoes you need for work are placed near the front door because you go to the office on weekdays, whereas the hiking boots stay buried deep in the closet until you pack for a yearly trekking holiday.

Ideally, data that is critical should be made available on expensive, quickly-accessed computer memory, or DRAM. While you may not put a price to the space near your front door, it is likely to be premium space that is kept free of clutter, so essential, often-used shoes are stored there. 

Data that is slightly less essential can be stored on the hard disk. A hard disk is not quite like a closet, but both are basically a cheaper alternative for storage with more capacity, and if well-organised, will allow you to get what you want quite quickly. 

Finally there will be items which you might want to place in long-term storage because they may never be used again, but which still have value. Data from a few years back that is archived to comply with legal requirements perhaps, or a baby's first booties if we continue with the footwear analogy.

In the IT world, applications can handle the categorisation, storage and retrieval of data, though data categories may need to be specified and budgets allocated for different types of data storage. While it may seem that DRAM is affordable and would be best for anything, it is still more expensive compared to disk-based alternatives, a fact which Ovum also points out.

Tiered storage options have existed for years. IBM and Oracle support federated data access, and backups were done to tape and optical media. What's changed is that today's enterprise software vendors have begun to champion the option in a big way, as not enough attention is paid to this aspect of data storage, and - done right - it offers the opportunity to save more money.

Tiered storage may well be the new black as Ovum says, and not just for analytic applications. The beginning of the year is always a good time for predictions. We'll see.