More and more solid-state drives based on flash memory are turning up in everyday products, signalling that the technology may be inching closer to mass adoption.

"SSDs are at an inflection point right now," says analyst Arun Taneja of the Taneja Group. "After years of incubation and false hopes, we have reached a point where there is adequate supply of reliable SSDs at low prices. ... There's not a single company in the storage industry that is not in some way, shape or form doing something with SSDs."

Storage vendors such as EMC and NetApp, and server vendors too, are raising the volume in their marketing of flash products. During this month alone, SSD product announcements have come from Texas Memory Systems, Sun, Dell, Pillar Data Systems, Compellent, Fusion-io and EMC.

STEC, which makes the flash memory used by many of the major storage vendors, announced two weeks ago that 2008 revenue for its ZeusIOPS SSD product line hit $53 million, three times more than in the previous year. Based on current demand, the 2008 total revenue figure will be topped in just the first six months of 2009, STEC says.

Moreover, Intel is shipping solid-state flash drives for use in servers, workstations and storage devices, and a company named SandForce will emerge from stealth mode in mid-April with a new SSD processor.

Flash memory is about 20 times more expensive per gigabyte than high-end Fibre Channel drives, EMC has noted. But huge advantages in both performance and power efficiency make solid-state worth it for applications that require a high number of IOPS (input/output operations per second). Today, many data centres buy up a ton of disk drives in order to improve speed, but end up with far more storage space than they need. Flash can deliver a high IOPS rate without wasting disk space, creating a cost-performance advantage, vendors say.

But that doesn't mean the current crop of flash-based products is ready for prime time. Burton Group analyst Gene Ruth says vendors are simply replacing hard disk drives with flash, rather than integrating SSD into the system in a way that takes full advantage of their high performance.

"Unfortunately, the choice of enterprise class storage systems with SSDs is pretty slim," Ruth wrote in a blog post on 12 March. "We need systems fully integrating SSD capabilities - not just replacing HDDs one for one. Replacing HDDs with SSDs is storage design 101. Easy to do but very suboptimal. Blending an SSD into a system is the real challenge."

What's happening today is that vendors are retrofitting existing products with SSD, allowing them to say "me too," Taneja notes. This is absolutely the right thing to do, according to Taneja, who says "you want to learn how to walk before you run."

But storage controllers on most existing products are not designed to handle both solid-state and magnetic drives, meaning that storage administrators have to decide which data should stay on the SSD layer and which should be moved to less-expensive storage, and then make that change manually. Ultimately, vendors need to build new products from the ground up that automatically move data from flash tiers to non-flash tiers based on changing needs, Taneja says.

Retrofitting has its benefits, "but the controller becomes the bottleneck," he says. "The only way to do totally innovative designs with SSDs is to say 'clean slate.'" Flash adoption in the enterprise is limited today, but should pick up once vendors unveil some truly innovative products, which should happen by 2010, Taneja says.