Driving SSDs Toward Greater Reliability

Arthur Cole

Pretty soon, it will be the rare enterprise that does not employ solid state storage in some fashion. Usually, the goal is to improve data throughput for high-value applications like Web transactions and database processing.

But just because you've added SSDs into your storage mix, does that mean you should trust them?

Reliability in solid state storage is an ongoing concern, and it looks like things will get worse as geometries shrink and densities increase. As Ars Technica's Lee Hutchinson shows us on a tour of the Flash structure, the benefits of smaller architectures — lower power consumption, increased capacity, among them — are outweighed by weakened gate structure, which makes it difficult to push electrons into and out of the cell at each write and rewrite. That means as Flash gets more efficient, it become more susceptible to wear and tear.

To counter this, drive manufacturers are turning to increasingly sophisticated technologies on the controller level and in management systems. Seagate, for example, recently acquired a stake in Israeli firm DensBits, which has developed the IP-based Memory Modem designed to improve data transfer in low-end NAND chips. Like most storage controller firms, DensBits keeps its technology under wraps, except to say that it pushes error correction and digital signal processing beyond current limits, much the same way that ADSL opened new doors in wide area data communications.

Meanwhile, firms like OCZ and LSI are looking to utilize their broad experience with storage technologies and data handling to improve SSD reliability. OCZ recently upgraded its Indilinx Everest platform with a new version of the Ndurance management system featuring a multi-level BCH ECC engine that pushes error correction to 128 bits per 1 KB, as well as proprietary adaptive programming that minimizes Flash deterioration through voltage shifting and advanced signal processing. At LSI, the new SandForce SF-2000 features advancements like 55-bit error correction per 512-byte sector, as well as the DuraWrite program cycle optimization system and the RAISE (Redundant Array of Independent Silicon Elements) protection and recovery system.

Probably the most intriguing development, however, is the launch of a company called Skyera, founded by former SandForce executives following that company's acquisition by LSI last year. The company claims to have developed a controller technology that boosts Flash endurance levels 100-fold, compared to the 10x results of current technology. Again, the company hasn't divulged many details of its system, other than that it provides "system-level error correction" that takes control of the physics of Flash storage.

In any age, the most advanced technology is only as valuable as its ability to maintain steady performance. The problem of late is that systems are becoming so sophisticated and are operating on such a micro scale that the slightest disruptions in data flow, temperature, vibration or other conditions can bring operations to a screeching halt.

Performance may be the benchmark for enterprise SSDs, but only as long as the device keeps running.

Add Comment      Leave a comment on this blog post

Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.