NEWPORT, R.I. -- Grid Strategies Inc. CEO Mike Bernhardt -- one of the most fervent advocates of grid computing in the IT industry -- will be the first one to tell you that there might not be a more controversial topic in the game right now than what both friend and foe alike have dubbed "the grid."
But Bernhardt is quick to argue that just because the idea of grid computing as a common enterprise solution is a divisive issue doesn't mean it's one standing on shaky ground. Keynoting a panel discussion titled "Grid Computing: The Convergence of HPC and Mainstream Computing" at the High Performance Computing and Communications Conference Tuesday, Bernhardt's message to attendees sitting on both sides of the grid fence was simple.
Like it, love it or hate it -- grid computing is here, and it's not going away.
But grid computing continues to fight a two-front battle against both perception and practicality. Even Bernhardt admits that, as little as a year ago, he believed in grid's bad reputation -- that the lack of standard definitions was slowing down its widespread adoption.
But this lack of a standard definition remains a major problem. Grid computing, according to a basic definition, is the application of the resources of many computers in a network to a single problem at the same time, and which requires the use of software that can divide and farm out pieces of a program to as many as several thousand computers -- essentially distributed, large-scale cluster computing.
But there is no one single grid, like there is no one Internet. However, the Web was launched with one set of standards, which led to widespread compatibility. Grid has had no such luck in finding such a common set of standards, leaving many to wonder how exactly to define what grid itself even is.
According to Wolfgang Gentzsch, the managing director of MCNC Grid Computing &Networking Services in Research Triangle Park, N.C., one of the first things users need to do when thinking of grid is to look past what it is and instead look at what it can do.
"Don't worry about definitions -- if it's distributed, connected by network, managed by middleware, it's a grid," Gentzsch said.
After seeing an increasing number of companies across a broader spectrum of industries adopt both grid standards and middleware, and following discussions he's had with HPCC attendees, Bernhardt said he's convinced the convergence of HPC and mainstream computing has picked up considerable speed in the past 12 months.
"What I'm finding is more validation. Folks that last year even were skeptical about grid, where it can go, what it can do and [if] it will it ever break out of the very tight niche high-end computing circle are starting to see there are more applications for it. I think the standards are helping to push that. They are fostering innovation," Bernhardt said.
According to Tony Lock, chief analyst at Great Britain's Bloor Research, the reason grid computing is gaining acceptance is because it is seen as a member the utility computing family. Utility computing has gained credibility in IT circles because it has potentially enormous cost benefits for companies.
"I think given recent rise in power of processors themselves coupled with Linux and grid infrastructure, more mainstream business applications are being deployed [on grid systems]," Lock said. "In that respect it's not completely bunk that HPC and mainstream are converging in areas where it makes sense … recent developments have made it more sensible and affordable in niche areas, financial services in particular."
Whether it's defined as a cluster or an advanced computing network, the capabilities of grid as a computing resource are undeniable. But translating that power into the adoption of grid technology as a mainstream enterprise solution remains to be seen.
Cliff Jervis, a partner in a Northeast-based computer company and an attendee at the HPCC conference, was one audience member who walked away from the discussion ready to give grid a chance.
"I'm excited for it. I think the key I see is that standards [are] a big deal. I don't think it's out there at all. Probably the largest obstacle from that comes from more from the politics of people accepting what it really means and how they are going to use it," Jervis said. "I would like the simplest user, someone with a browser sitting in an Internet cafÉ in Brazil, able to hook into somebody in New York City. With those kinds of connections, local and global, I think the grid could work for that with standards-based type tools that anyone could use."
Let us know what you think about the story; e-mail: Luke Meredith, News Writer