Intel, Micron to double SSD density by summer

Micron plans to ship a new enterprise-class PCIe solid state drive this summer

SANTA CLARA, CALIF. - Intel and Micron plan to further shrink NAND circuitry, doubling the density of their flash chips and further reducing the cost of solid state drive (SSD) storage.

IM Flash Technologies (IMFT), Intel and Micron's joint venture, released 25 nanometer (nm) circuitry a little over a year ago.

In the next few weeks, IMFT plans to annouce 20nm NAND flash chip production.

Early this summer, the company plans to release an enterprise-class SSD based on the PCIe expansion card standard, but it won't be based on 20nm NAND. The new offering will be based on older 34nm NAND process technology.

"This will be the industry's leading drive," Kevin Kilbuck, Micron's director of strategic marketing for Micron's NAND Product Group, told Computerworld at the Storage Networking World conference here this week.

The new PCIe card, called the P320h, follows Micron's first enterprise-class SSD, the P300, which is based on the serial ATA (SATA) interface and single-level cell (SLC) NAND flash memory.

The P320h will also use SLC NAND flash along with a technology based on a familiar acronym with a different twist: RAIN, or redundant array of independent NAND. RAIN is more commonly defined as "redundant array of independent (or inexpensive) nodes," which refers to the building blocks of a grid storage architecture that incorporates both processors and disk storage in one unit.

In Micron's instance, RAIN will refer to utilizing additional NAND as a cache or buffer to increase resilience.

Unlike Intel, which last week said it was as much as walking away from SLC, Micron sees a future in offering SLC-based SSDs as the highest tier of storage in an enterprise, even if that tier only represents about 10% of SSD capacity sold.

Kilbuck contends that 10% of capacity will represent about 50% of market revenue by 2014.

In contrast, Intel said its most popular consumer-class SSD, the X25-M, outsold its SLC-based enterprsie-class SSD, the X25-E, by as much as 7:1 to 8:1 in the enterprise. So in the future the company plans to focus on enterprise-class products based on MLC NAND.

Micron admits its consumer-class SSD, the C400, is also far outselling its enterprise-class SSD, the P300. But, in terms of revenue, it expects enterprise-class SSDs to match that of the less costly consumer products.

For example, Micron separates enterprise SSD into three categories:

  • Enterprise performance products based on PCIe and SAS interfaces that are aimed at OLTP, high performance computing and relational database applications.
  • Enterprise mainstream products based on SAS and SATA interfaces that are aimed at web servers and data caching.
  • Enterprise value products based on the SATA interface and aimed at being a boot drive, storage for log files in databases and data vaulting, or a DRAM offload point.

Kevin Dilbelius, Micron's senior product marketing manager for enterprise SSD, said the cloud is driving the need for higher performance flash products that can help blade server infrastructures serve up things like online streaming video.

IMFT is coming up against considerable technology hurdles in its attempts to continue reducing the size of NAND flash memory. At 25nm, IMFT is approaching atomic sizes. For example, a human hair is 3,000 times thicker than 25nm. If the thickness of a human hair was one mile, 25nm would be about 20 inches, according to Kilbuck.

At the 20nm level, Kilbuck said only a handful of companies will have the intellectual property that will allow them to continue producing SSD products.

"When we were at 50nm, pretty much anyone could slap together an SSD," he said. "At 20nm, it's almost the opposite effect."

There are inherent problems in shrinking the size of circuitry used in semiconductors, most notably an increase in data error rates from electrons bleeding through ever-thinner silicon walls. That requires the development of more sophisticated error correction code (ECC) and signal processing algorithms, as well as overprovisioning NAND flash chips to guard against data loss.

"We see a big sea change coming in the SSD world where third-party vendors simply don't have the IP to play," Dilbelius said.

Lucas Mearian covers storage, disaster recovery and business continuity, financial services infrastructure and health care IT for Computerworld. Follow Lucas on Twitter at @lucasmearian, or subscribe to Lucas's RSS feed . His e-mail address is lmearian@computerworld.com.

Copyright © 2011 IDG Communications, Inc.

Bing’s AI chatbot came to work for me. I had to fire it.
Shop Tech Products at Amazon