I got to spend a couple of days at the TheServerSide Java Symposium last week, and I was rather impressed with the mindshare grid computing was given at the event. While I had never before been to the conference, most everyone to whom I spoke echoed this sentiment, and it was illustrated in the tiny exhibition area, where the GigaSpaces and Appistry booths, along with Google, comprised a a rather large fraction of the exhibitors. And grid’s presence wasn’t limited to the booth area.
There were several sessions that fell under the “grid” heading, and it seemed like GigaSpaces founder and CTO Nati Shalom was everywhere, as was Tangosol CEO Cameron Purdy (more about Tangosol later). Also flying the grid flag participating in various capacities were Appistry CTO Bob Lozano and representatives from IONA and Terracotta. Of course, none of the aformentioned companies play in the traditional grid space, they tend to fall under what Gartner calls the grid-based application platform market. And this was made quite clear in a Thursday-afternoon panel entitled “High-Performance Architectures.” (On a side note, this panel got off to a rather uncomfortable start when the moderator, IONA Technical Director John Davies, made each participant choose a bottle of beer to drink throughout the discussion. I think the four panelists combined to drink a single bottle, hardly enough to facilitate a loose-lipped and controversial discussion of Java-based application platforms — if there is such a thing. Gotta give Davies points for trying, though …)
Back to the point: Not surprisingly given the panelists — Shalom, Lozano, Gil Tene of Azul Systems and independent consultant Kirk Pepperdine — the term “high-performance” was discussed not in the batch-oriented, CPU- intensive sense of traditional HPC and grid computing, but rather in the light of extreme transaction processing and the ability for computing resources to have access to data in an event-driven manner. To me, it signified yet another sign that the world is changing, and companies looking to follow the leads of Google, eBay and the rest of the Web 2.0 giants will need architectures that allow them to do so.
Which brings us back to Tangosol and the announcement Friday morning that Oracle is set to acquire the Somerville, Mass.-based company. This is HUGE news in the grid-based application platform space, and seemed to cast a shadow over much of the day’s goings on at the conference. Heck, Purdy even sported an Oracle shirt during a lunchtime panel and coined, I believe, the word “Orasol.” The common take on this acquisition seems to be that it is a validation of this market and, along with a number of other acquisition over the past few years, a concession from Oracle that the traditional database is on its way out. I haven’t heard much in terms of details, such as what Purdy’s role will be once the acquisition goes through, but I did hear rumors that the purchase price was somewhere between $60 and $80 million. No wonder even Tangosol’s competitors are offering congratulations. We’ll have more on this story next week, including comments from these competitors, as well as the analyst community and, hopefully, the two companies involved.
As for this week’s issue, be sure to read the article on GXS’ Trading Grid, which provides yet another take on the concept of on-demand computing. A combination of on-demand computing, utility computing and SOA, it certainly is a unique solution. And with 41,000 customers, GXS must be doing something right. I’d also like to point directly to “The Research Network-Grid Link,” which provides a background on the relationship between high-performance research networks and grid computing — something that should be of interest to many in the community. On top of that, we have Larry Smarr getting honored; United Devices making more pharma news (click here and here); Appistry and Tangosol announcing product updates; PRAGMA and friends joining the fight against avian flu; and the inaugural German e-Science Conference, which was brought to my attention by our good friend Wolfgang Gentzsch. And there is much, much more, so take a close look at the table of contents.
I’d like to close this week by commenting on the countless blogs I’ve read in the past few days citing the enormous power increase Folding@Home has experienced since the PlayStation 3’s got on board. As of Friday, there were about 13,000 PS3s contributing to the project, and their Cell processors brought with them 338 teraflops, more than doubling the project’s output to 579 teraflops. Since then, I’ve seen sources putting that total number over 900 teraflops; it would appear a petaflop of power for Folding@Home is just around the corner. Maybe it’s already there.
While many have dismissed the cycle-scavenging scene as simplistic and a poor representative of the totality of what grid computing can do, Sony and the PS3 seem to have breathed some life into it. Not only will this news likely at least improve video games’ reputation, but it also might stand to shake up the distributed computing and processor markets. I can’t be alone in wondering why a video game console makes my PC seem no more powerful than a calculator.