IBM leaps two hurdles for next-gen memory
IBM has solved two related problems with phase-change memory and now says the next-generation data-storage technology will be ready for use in 2016 in servers. — In a paper for the IEEE International Memory Workshop, Big Blue researchers describe …
Context & Ripple Effects
IBM researchers reported at the IEEE International Memory Workshop that they had cleared two related technical problems with phase-change memory, and set 2016 as the year the technology would be ready inside servers — a rare public timetable for a lab-stage storage material.
First-order effects
- IBM now has a credible candidate to sit between DRAM and flash in its own server line, giving Big Blue's systems business a five-year internal target rather than a vague research direction.
- The announcement puts incumbent DRAM and flash makers on notice that a deep-pocketed rival is aiming directly at the price-performance boundary their two product lines share.
Second-order effects
- Server buyers gain the prospect of a third storage tier, changing the capacity-versus-latency trade-off that currently forces them to overbuy fast memory or accept slower flash.
- Competing memory vendors face pressure to accelerate their own next-generation programs, since a working PCM supply aimed at 2016 servers would erode the pricing power flash holds in the data center.
Third-order effects
- If PCM matures on IBM's timetable, the hard wall separating 'memory' from 'storage' in server architecture starts to dissolve, reshaping how systems vendors design tiers and how memory markets are priced.
The trend: Data-center storage is moving toward a blurred memory hierarchy where single technologies promise DRAM-like speed at closer-to-flash cost, with large systems vendors like IBM funding the transition in-house.