Subscribe

Not just a flash in the pan

Flash will be an important part of the storage mix in years to come, but it won't replace spinning discs any time soon.

Lance Harris
By Lance Harris, freelancer
Johannesburg, 01 Jul 2013

Big data, the cloud and high-end analytics are all helping to spur demand for flash-based storage arrays in the enterprise. But most industry observers agree that the prohibitive costs of the technology will limit its deployment to a few select environments and applications where speed really matters.

Although it's a much smaller market, we do see increasing use cases for flash arrays.

A flash storage array is a solid state storage system that contains multiple flash memory drives instead of hard disk drives. Flash first started to find its way into hybrid flash/disc arrays provided by the major storage vendors as far back as 2008, complementing the hard drives used to store most of the data.

Since then, the hype has gone into overdrive. However, the consensus is that flash has its place, but is not the right solution for every environment.

Benefits of the technology include its lightning-quick speed and low latency. It allows data to be accessed and transferred much faster than electro-mechanical disc drives, offering a major advantage in speed for frequently accessed application data. But this comes with a hefty price tag.

Complexity

Even today, flash costs around seven times as much as spinning disc, says John Rollason, director of product, solutions and alliances marketing for EMEA at NetApp. That means hybrid solutions that combine flash and hard disc drives are likely to dominate the market for the foreseeable future.

"Although it's a much smaller market, we do see increasing use cases for flash arrays," says Rollason. "A lot of Web applications need very low levels of latency, for example, and hyper-scale computing providers use flash." The flipside is that there are a lot of environments that don't need flash or need very little, he adds.

To optimise storage solutions for performance, organisations are placing their most frequently accessed application data (or hot data) on high-performing solid state storage and less frequently accessed data on capacity-efficient hard disc.

As the complexity and cost of corporate data storage requirements continues to increase, there's no one-size-fits-all solution, says Graeme O'Driscoll, innovation and technology manager for cloud at Internet Solutions.

Many companies are storing large amounts of data online using tiered storage - a combination of solid state drives, fibre channel, serial attached SCSI and serial ATA (SATA) drives. This approach drives down the per gigabyte costs, as data can move down tiers based on storage duration and access requirements.


A fully automated intelligent storage tiering system monitors data blocks and identifies which are the 'hottest'. The system moves these to SSD storage, and colder blocks to SATA disks. This delivers a lower blended price, while ensuring performance remains high for key data that needs to be accessed fast.

Share