Start-up Pliant claims to have secret for speedier SSDs
Flash-based storage with RAM-based IOPS
Fledgling storage firm Pliant Technologies has thrown off its startup invisibility cloak, and claims it can greatly improve the performance of Flash-based solid state drives.
Pliant said this week it received $8m in first round funding funding, led by Lightspeed Venture Partners, to develop a "new class" of SSDs for big business data centers.
The secret sauce is apparently in the controller, although Pliant won't issue any meaty specifics until it gets closer to a planned Q4 product release. That said, the company has an impressive enough management team to take it seriously. You'll find chairman Jim McCoy, co-founder of Quantum and Maxtor, and CEO Amyl Ahola, former chief of TeraStor and veep at Seagate.
Ahola told El Reg that the drives will use a proprietary controller that lets a flash-based SSD come close to matching the speed performance of Dynamic-Random Access Memory (DRAM) SSDs.
"We use standard, off the shelf flash chips with an architecture and software for the controller that's unlike anything else in the market," said Ahola.
In general, RAM-based SSDs enjoy greater input/output operations per second (IOPS) than their flash counterparts. But Flash SSDs have a physical size and capacity advantage — as well as not requiring emergency backup power to prevent data loss.
But traditional flash SSDs leave IOPS crunching systems wanting. A high end Flash-based drive operates at around 25,000 IOPS. A good DRAM drive can do 400,000. Ahola isn't giving figures yet, but claims Pliant's flash SSDs will be much closer to the latter figure. And the products will come in the form factor of a traditional drive.
Ahola estimates the controller will make up about one or two per cent of the cost of the unit.
"Because of the architecture that we use, the drive doesn't add any significant cost over traditional high-end flash drives," Ahola said.
That's not to say you should start counting your pennies. High performance flash-based SSDs are an investment even for giant, high rolling data centers.
But Ahola argues that flash storage could save money for large companies. He said that corporate data centers often purchase three to four times the amount of bleeding-edge disk drives that they really need to get the best IOPs performance. Switching to flash storage can eliminate over-provisioning — and therefore save rack space, reduce power and cooling and all that other green computing hullabaloo.
Ahola said the first round of funding will allow Pliant to complete its development efforts that began two years ago and start preproduction of product.
"It will last us through the end of the year," he said.
The first test units will begin popping up around the summer and fall. Pliant plans to sell the drives through OEMs. And yes, the identities of those vendors are secret too. ®
re: Consumer Market?
Are you actually saying that you consider a 4GB Sandisk/PNY/whatever USB flash-based memory stick to be a "high-end flash drive"? If so, then please cover yourself in foil so we can identify you. By "high-end flash drive", he's talking about actual SSDs, not flash-based memory sticks. I would assume he means something with 128GB+ capacity, and something with an actual drive interface (FC, SAS, SCSI, SATA, etc), which currently cost $3000+.
Patent application from the founders
A patent application from two co-founders of Pliant, Olbrich (CTO) and Prins (Chief Architect), can be found by Googling for USPTO Application #: 20070294468 "Architecture for reading and writing to flash memory"
Not exactly earth shattering - use of multiple processors between a DDR RAM buffer and flash devices. [sarcasm] I wonder why that hasn't been thought of before?[/sarcasm]
yes, but one can balance reads and, especially, writes, across multiple chips. The more chips you have and the better logic to balance writes, the bettter potential performance improvement.