March 24, 2008
I had an interesting conversation this week with Forrester’s Frank Gillett about his latest report, which concludes that the term “grid computing” (as well as “high-performance computing”) does not generate interest in typical enterprises. The report includes all sorts of percentages regarding levels of interest and adoption, but the bottom line (at least in North America and Europe) is quite simply that there are more IT decision-makers not interested in grid computing than there are IT decision-makers interested in grid computing.
Gillett attributes much of this apathy to the fact that grid computing is, in general, beneficial to vertically specific applications. Most businesses, he notes, do not have much use for its ability to run parallelized, compute-intensive workloads in their general-purpose environments. “There will be usefulness in general-purpose solutions like Platform and DataSynapse,” he explained. “However, with any application that benefits from [grid computing] in a big way, the ISV will build it in, or they will OEM the technology from one of those guys.” In some cases, he added, organizations might be taking advantage of grid computing without realizing it because it has been embedded into their applications out of the box.
Alternatively, I think much of this disinterest might result from people applying such a constrained definition to “grid computing.” Grid has spawned several new technologies over the past few years (e.g., grid-based application platforms, applications fabrics, etc.) that, if considered grid computing, seem applicable to a far broader range to users. When asked just about “grid computing,” though, with no explanation of what that encompasses, respondents go with what they believe to be the reality. Clearly, however, enterprises do not perceive these related solutions to be grid computing, so perhaps it is my definition that needs to be amended.
And in this disparity of definitions lies the gist of Frank’s report, and the problem with the grid market, in general: When it comes to enterprise IT mindshare, “grid” has been bandied about so much that no one knows what it means or what business benefits they might derive from it.
For example, some vendors have pitched grid computing based on its ability to consolidate resources and increase utilization, which has led some users to conclude that x86-server virtualization, which offers those same benefits, solves the same problems as grid computing. As Gillett points out, though, that’s simply not the case because while server virtualization achieves high utilization by allowing one server to do several things concurrently, grid computing achieves high utilization by letting users allocate multiple servers onto a single task. “When we look at grid computing, what we see is a technology that, in some ways, makes it easier to stand up a bunch of virtual servers to act as grid nodes and then tear them down again and repurpose the machines for something else,” he told me. “To me, they’re not at all in conflict, nor are they competitive. They are, in fact, complementary.” He added that while virtualization might have won the marketing battle in terms of which word sells software, we must consider the technology battle a “no contest,” as grid computing and virtualization are different technologies.
On an interesting side note, Gillett and Forrester are predicting that “virtualization,” too, will burn out as a marketing term because of the confusion that comes along with its rapidly expanding usage. A prime example of this expanding usage is the term “application virtualization,” which has at least two distinct meanings, one of which sometimes is used interchangeably with “grid computing.” It shouldn’t surprise anyone that this has led to some confusion, and when you throw into the mix I/O virtualization, storage virtualization and network virtualization (which isn’t actually the virtualization of the network), people start to lose track of which way is up.
Because users haven’t yet been willing to educate themselves on the differences between various types of grid set-ups (e.g., Gillett’s defined categories of loosely coupled grids, tightly coupled grids and uncoupled grids) and various types of virtualization, as well as the varying business drivers unique to each one, vendors are left struggling to define what they’re offering. Analyst firms have coined terms like “organic IT” (Forrester) and “real-time infrastructure” (Gartner), but neither, says Gillett, is the right mix of descriptive and evocative to really catch on -- at least not yet. When vendors like DataSynapse attempt transition from selling grid computing to selling application virtualization to selling real-time infrastructure, Gillett said they’re taking steps in the right direction, but no one has yet to find the right pitch. Obviously, he noted, DataSynapse could go into sales situations telling customers “we accelerate your application delivery platform” or asking whether customers are having trouble scaling out a parallelizable Java application, but these IT-level phrases lack an evocative business element.
Encapsulating the entire tone of our call into two sentences, Gillett laid out why this naming problem is such an issue. “We have a bunch of complex ideas, products and technologies whose application is specific to certain situations, and we’re trying to talk about it in generalizable terms because, frankly, most people can’t wrap their heads around all the details,” he said. “I can sit down and in an hour sort of go through the landscape with someone, but that’s a terrible way to sell a product.”
One of Forrester’s clients, a “large hardware systems” vendor, already has taken the firm’s advice and throttled back the grid talk as a major emphasis of its product, as have several smaller vendors, but even that, said Gillett, isn’t a cure-all. As vendors try to sell in terms with which users can relate, users -- in both the IT and general business departments -- must really begin to understand what they need and how to describe it. Quite honestly, though, I don’t know that we will see this meeting of the minds come anytime soon because, as Gillett alluded to, the level of complexity stemming from the often-minute differences between platforms and their varying business benefits is a lot to overcome. “None of us … have figured out what the simple way to talk about this is,” he concluded, “because it isn’t simple.”
Speaking of using grid and virtualization interchangeably all while trying to highlight business benefits, be sure to check out this week’s feature article on managed hosting. It looks at two more companies -- Radiant Communications and Mosso -- that have jumped upon the virtualized grid bandwagon and are selling these VMs, as well as their management, as hosted services. The business drivers, of course, are the speed and ease with which customers can scale their resources, as well as the ability to offload maintenance costs, etc., to a service provider. Given all the talk this year about utility computing, it is worth your while to keep up with who’s who and who is offering what.
Elsewhere in the issue, there are a lot of big announcements, including: the latest news on IBM’s cloud computing push; 3Tera and Nirvanix introducing a “cloud storage” solution; Microsoft and Intel’s partnership around parallel computing; Allegro’s grid-ready energy trading application; an advancement in blade virtualization; and EMC’s fabric backbone. And, if you follow the trials and tribulations of the Open Grid Forum, be sure to read our Q&A with the OGF regarding its recent insights into who its stakeholders are and how this will affect its priorities going forward.
Posted by Derrick Harris - March 24, 2008 @ 11:46 AM, Pacific Daylight Time
Derrick Harris is the Editor of On-Demand Enterprise
No Recent Blog Comments
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?