In his 1956 dystopian sci-fi novel "The City and the Stars", Arthur C. Clarke puts forth the fundamental design tenet for making eternal machines, "A machine shall have no moving parts".
To someone from the 1950s current computers would appear to come close to that ideal - the CPUs and memory perform silent magic and can, with some ingenuity, be passively cooled, and invisible electronic signals carry information in and out of them to networks and oops, to rotating disks, still with us after more than five decades[i].
But, as we all know, salvation has appeared on the horizon in the form of solid-state storage, so called flash storage (actually an idea of several decades standing as well, just not affordable until recently).
The initial substitution of flash for conventional storage yields immediate gratification in the form of lower power, maybe lower cost if used effectively, and higher performance, but the ripple effect benefits of flash can be even more pervasive. However, the implementation of the major architectural changes engendered across the whole IT stack by the use of flash is a difficult conceptual challenge for users and largely addressed only piecemeal by most vendors. Enter IBM and its Flashahead initiative.
What is Happening?
On Friday, April 11, IBM announced a major initiative, to the tune of a spending commitment of $1B, to accelerate the use of flash technology by means of three major programs:
- Fundamental flash R&D
- New storage products built on flash-only memory technology
- An initial set of 12 flash centres of excellence. In these centres IBM will work with customers to explore the optimal application of flash for a user's specific problems and environments, develop reusable flash IP, and provide other flash-related services.
This initiative is largely based on the technology acquired from Texas Memory systems, possibly one of the smarter acquisitions IBM has made in recent history. TMS was a mature company, more that 30 years old with a significant portfolio of flash-related patents and a very attractive customer base, concentrated in Oracle DB users and a strong customer footprint in defense-related government, healthcare and telecom sectors.
The Economics of Flash - Look at the System, not the Device Level
One of the cornerstones of the IBM announcement is a message that has been promoted by others in the recent past, but none with such a big soapbox as IBM - that even unoptimised (no compression or deduplication) flash is cheaper than disk as it is deployed in typical database environments, primarily because most high-performance database environments use disk very inefficiently, with multiple spindles used in parallel for performance with poor utilisation.
Flash, with its inherently low latency regardless of the physical location of the data, in the end nets out at around a claimed $10/GB compared to $10 - 12/GB for an equivalent high-performance disk layout in a typical database environment.
Couple the fundamental efficiencies of flash with the prospect of continued improvements in flash density and pricing and the ripple effect of improved latency on overall system performance and the opportunity to reduce power costs, space consumption and even reduction in per-core software license fees as server resources become more efficiently used, and the economics of flash look very bright.
What is needed to catalyse the widespread adoption of flash is big names to push flash to the masses in the form of education, sales channels and products. Established storage vendors are beginning to do so, along with other systems vendors, but IBM's whole-hearted endorsement and high-profile support raises the bar in a major way.
What makes this different from other programs?
The difference is that IBM is one of the few organisations with the perspective and resources to see across the entire enterprise technology stack, from software through the supporting infrastructure, and has the resources to build and deliver a comprehensive program like this. Even in the presentation of detailed cost models at the unveiling of this initiative, IBM displayed a degree of sophistication in understanding the flash vs. disk cost and value model that is often missing in other vendor presentations.
My impression is that this is a program whose timing is correct, applied to a technology that has achieved sufficient maturity, and addressing a very real set of problems plaguing many IT shops today - the immense proliferation of rotating disk spindles and their accompanying cost, power consumption nd space requirements.
Many of the elements of IBM's programs are probably not unique, but IBM's proven ability to structure complex solutions and deliver them to customers will, I believe, serve them and their customers well, and will no doubt be an inspiration to their competitors. In the end, IT consumers across the industry will benefit.
[Note: My colleague Henry Baltazar, who has written and will continue to write on flash memory, was a major contributor to this blog post.]
[i] Lest the real purists scold me, some of the very first digital computers did indeed have various forms of persistent memory that did not involve moving parts.
Posted by Richard Fichera