SARATOGA, Calif. – Seeking to propel solid-state storage (SSD) drives into more of a mainstream product, SandForce Inc. has rolled out its latest controllers in the arena.
Based on a 65-nm process from its foundry vendor-Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC)-the SF-2000 family of SSD controllers or processors from SandForce (Saratoga) include three parts. The SF-2300 is aimed for industrial applications, while the SF-2500 (SATA) and SF-2600 (SAS) are geared for the enterprise.
The company is leveraging the technology from its first products, based on a 90-nm process from TSMC. Last year, as part of a major rollout, SandForce claimed to be the first company with an SSD processor supporting multi-level cell flash (MLC) chips.
SandForce is one of several companies chasing after the emerging SSD merchant controller market. Indilinx, Jmicron, Marvell and others are also in the game. Several SSD players make their own controllers, including Samsung, SanDisk, Toshiba, Intel and Micron. Intel and Marvell reportedly have collaborated on an SSD controller, sources said.
Thanks to its proprietary reliability and related technologies, SandForce is one step ahead in the overall SSD controller market, said Kent Smith, senior director of product marketing for the startup.
Still to be seen, however, is if or when SSDs will take off. SSDs are NAND-based storage devices that claim to be lower power and faster than traditional hard disk drives (HDDs). But SSDs have not taken off as fast as previously thought, due to cost and reliability issues.
SSDs have seen some success in netbooks, notebook PCs and other products. The enterprise is just ''starting now’’ to see the adoption of SSDs, Smith said.
SSD prices are only part of the equation. The SSD itself is managed by what some call an intelligent controller. As storage functions become more complex, SSD controllers are becoming critical, he said.
The question is whether OEMs devise their own controllers or buy a merchant solution. For now, ''Samsung uses a proprietary controller (for their SSDs). So do SanDisk, Toshiba, Intel, and Micron,’’ said Jim Handy, an analyst with Objective-Analysis.
There could be a sea of change in the market. ''The cost of developing a controller is so great-and with a very high risk of failure- that I anticipate more and more companies will migrate away from proprietary controllers, especially in client SSDs, which are those aimed at the PC rather than the enterprise,’’ Handy said.
However, there is no clear-cut winner in the emerging merchant SSD controller space. ''The (merchant) controllers from the outside are good,’’ he said. ''I expect to see this industry have a ‘preferred controller du jour’ with each controller vendor enjoying a day in the sun before another company develops something better and takes the lead away.’’
SandForce could be leading the field by a nose in some respects. ''They are not sitting still,’’ Handy said. ''They have bested their own high-performance chip before their competition could even catch up. SandForce appears to be intent on keeping the speed title to themselves. This should help them grow their market share.’’
Founded in 2006, SandForce last year closed $21 million in series C funding. Led by investor TransLink Capital, the round also included new investors UMC Capital, LSI, Red Maple Ventures, Darwin Ventures, and A-Data Technology as well as all of the existing SandForce investors--DCM, Storm Ventures, and tier-1 storage OEMs.
Last month, SandForce announced that it had closed $25 million in Series D funding, led by Canaan Partners. Existing investors who also participated include DCM, Storm Ventures, Translink Capital, LSI and UMC Capital.
SandForce is looking to help bring SSDs into the mainstream. This time, the company is attacking the industrial and enterprise.
To date, flash drives for servers have mainly used single-level cell (SLC) flash chips. But SSDs based on SLC are expensive, prompting the need for MLC-based products. But the endurance and reliability for MLC chips have generally not been up to the requirements of servers.
NAND flash memory has a limited number of program-erase (P/E) cycles. Older NAND products are said to withstand around 100,000 P/E cycles. Then, wear begins to deteriorate the reliability of the device. The newer and higher-density NAND parts have far fewer P/E cycles. Thus, NAND reliability becomes an issue in SSD designs.
SandForce claims that its SSD processors can reduce the cost of MLC-based SSDs, while maintaining enterprise reliability and duty cycles. As part of the latest effort, the company rolled out the SF-2000 lineup.
These chips feature a 6-gigabit-per-second SATA host interface, 60,000 sustained random read/write IOPS (input-output operations per second), and sequential read/write performance of 500-megabytes-per second.
To solve the reliability and other issues, SandForce’s SSD processors include so-called Raise technology, which reduces field failures and returns. Another technology, DuraWrite, optimizes MLC endurance in write intensive applications.
In addition to Serial-Attached SCSI (SAS) connectivity enhancements, the SF-2000 family supports single-level, multi-level, and enterprise multi-level cell NAND flash families from Hynix, Intel, Micron, Samsung, and Toshiba. It supports the ONFi2 and Toggle Flash interfaces with data rates up to 166 mega transfers per second
These devices reduce cost, as they do not use DRAM. This is optimal for SATA-based SSDs, in addition to SAS- and PCI Express-based SSDs when integrated with RAID and host-bus adapter chips on one card.
The chips include dual-ported SAS bridge support, including non-512-byte sector sizes. It also includes an ECC engine correcting up to 55 bits per 512-byte sector. And they also support TCG Enterprise security with selectable multi-banded 256/128-bit AES encryption with line-rate double encryption for data written to the drive.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.