SAN JOSE, Calif. In an effort to regain its old edge as a leader in computer storage, the company that invented the hard-disk drive used the technology's 50th anniversary to sketch out its advancements in phase-change RAM, tape drives and disk arrays. In doing so, IBM Corp. also aimed to pave the way for the launch of its IceCube effort as a novel class of storage array.
IBM has partnered with Infineon spinout Qimonda AG and Taiwan's Macronix International Co. Ltd. on phase-change RAM, the company said in a briefing at its Almaden Research Center, home to much of the pioneering work in disk drives. An upcoming technical paper from the trio will report success on the development of devices with features as small as 10 nanometers, a key development in an area where competing magnetic RAM is weak.
Separately, the company said it is on track to commercialize a new class of tape drives with significantly boosted capacity, thanks to the first use of arrays of giant magnetoresistive heads. And IBM re- searchers are developing systems software to expand the capacity and computing clout of disk arrays that handle big pools of storage in the data center.
Despite the announcements, "They are still really afraid of the folks in Hopkinton, Mass.," whispered James Porter, a veteran storage analyst who attended the briefing. Porter was referring to EMC Corp., an IBM archrival in data center storage systems.
Hard drives are still the focus for mainstream storage, though they constitute a consolidating, commodity market. After spinning out its hard drive business to Hitachi, IBM now plays only indirectly in that market as a vendor of large hard drive array systems that sit in corporate data centers. Even in the data center, IBM has had to acknowledge innovators like Network Appliance Inc., whose highly successful network-attached storage systems IBM resells under its own brand.
Conveniently, the future of storage technology may not be in the hard drive, according to IBM researchers. As disk heads approach tens of nanometers in size, progress in drive capacity and bandwidth is slowing from 60 to 100 percent a year to less than 35 percent a year, even as the amount of storage shipped each year grows by 50 percent or more, said Dilip Kandlur, director of storage systems research at the Almaden facility.
Thus IBM is pursuing research in three major areas. It seeks chip-based alternatives to drives with active programs in phase-change RAM and MRAM. It's pushing the frontiers in tape storage, which it expects will have a long life in archival systems. And it's developing software to bring more computing functions to storage arrays.
IBM researchers hope to find a semiconductor memory to replace flash and disks. It will need to have a small cell, low power budget, low cost, 100-Mbyte/second transfer rates, support for 1012 read/write cycles and a 10-year life cycle.
They believe phase-change RAM holds the most promise. A joint paper from IBM, Qimonda and Macronix at this year's IEDM conference will report scaling advances that have IBM research manager Gian-Luca Bona excited, although skeptical. "We still have to put all the pieces together in prototypes," said Bona.