Grid Spending Today

GCJ: What I'd like to do is first have you tell us about the report, and some of the details.

Rosenberg: Insight Research started looking at the segment in 2003. 2003 grid was still very much a research-oriented capability. We're primarily telecommunications analysts; we're not IT jocks. But the implications were that this technology, the nature of virtualization, and what it will mean, would substantially alter our clients' - that is, the telecommunications companies - requirements.

Back in 2003, the telecommunications industry was in the potty. And to a certain extent it is still suffering from overbuild. Essentially, there was too much capacity in the backbone networks. In 2001, 2002, 2003 it wasn't obvious that there were traditional telecom applications on the horizon that would fill those pipes. And the interest was piqued in grid because of the nature of the application, and what it might mean to that - essentially taking up some of the slack. Start to use capacity in a meaningful way to pull the telecommunications industry out of the doldrums.

Now, as it turns out, it looks like a lot of that capacity will be absorbed more quickly by the phone companies' entry into distribution of video services - in other words, TV, video telephony, movie distribution- and that sort of thing. In the meanwhile, grid is - as we said in the 2006 study, it's progressing a little bit slower than we anticipated at the time.

We're still very keen on the technology - that is, the benefits of it are clear. The nomenclature that's used, whether it's grid, or service-oriented architecture, or virtualization - data center virtualization - kind of muddies the waters a little bit. We're sticking with the standard industry usage and very specifically define the grid space as opposed to tangential areas that are, in some of the less rigorous journals, sometimes associated with grid.

GCJ: Question on the methodology for the report. How did this all come about? Did you do interviews, or how did it happen?

Rosenberg: All of Insight's market research reports are done through a triangulation method. I guess that is the best way to describe it. You have to assemble data points in a number of different ways. One way, as you said, is to interview a number of industry experts, and contact the major players. And it's really not a tremendously onerous job, because if you look at our table of contents - I just want to count the number of vendors we've actually written up. Looks like - one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen. You know, more than a dozen, a little bit more than a dozen vendors who are making the market. So it's not an onerous task from that point of view.

The other point of data collection, of course, would be to look at the industry adoption rates, and talk to those that are building the systems - go to conferences and talk to people who are actively involved in the implementation.

And then the third data collection mechanism is to essentially scan the literature, keep on top of literature. So all of those methodologies of data collection coalesce in the final report.

Now, what we've spent time doing in this study is to try and understand the dynamic of grid by vertical industry - which industries are moving faster than others. Because it's not going to be uniform, new technology adoption is never uniform. There are certain industries that have a higher propensity to adopt the newest developments in IT because the return to them is significantly higher than - in other, a different industry and a different segment.

So we all know that financial services has been an early adopter, advanced engineering, airframe manufacturing has been an adopter, pharmaceutical has been an adopter. And what we saw, again, going back to our particular industry, our particular way of coming at this subject area was we've seen a number of telecom companies now start to wet their feet - DT and Telefonica both selected middleware partners to build service delivery capabilities.

GCJ: What is the characterization of an early adopter?

Rosenberg: Typically what we saw in these early-adopter implementations was either a department would adopt it for a particular application space, or an entire location would adopt it to improve its ROI on its investment resources. But they were typically kept behind the firewall. So there could be substantially monetary investments, but the monetary investment was either at a specific location or within a specific group. And typically did not span an entire enterprise

For example, in the financial services industry some outfits were doing very sophisticated market modeling of derivatives- which are very complex financial instruments-using grid applications. But it wasn't spread across the entire bank. It was specifically in an application area that could be - and readily - and that's what I'm talking about what I say when I say early adoption phase.

GCJ: You're sizing how much is going to be spent in this space, is that correct?

Rosenberg: We've created a market profile for spending on grids by a vertical industry. So we looked at about 12 or 14 different vertical industries. We looked at healthcare - and this will all be in the press release. Healthcare, construction, retail trade, wholesale, education, social service, financial insurance and real estate, professional business services, transportation communications, durable, non-durable manufacturing, etc. So we tried to look at the grid adoption - or spending for grid tools by the different verticals.

GCJ: Let's talk a little bit about usage, from a compute grid or a data grid or an app grid. What are folks actually using out in the field, the folks that you're talking to?

Rosenberg: Worldwide right now, largest spend is an enterprise. It's within the enterprise - it's not a partner, it's not a service. In this order: Compute, data, application, then instrumentation.

GCJ: What's your take on Grid/Virtualization/SOA, based on what you found in your research?

Rosenberg: I think that the clear thing is that we're moving, finally, to a different paradigm of computing. Grid is fairly well-articulated in the literature as to what constitutes a grid. As a matter of fact, I was on Wikipedia the other day, and I was looking at the definitions that were applied. We follow those pretty carefully in order to constrain ourselves so that we're not talking about areas congruent to the subject, but not on point.

But again, I think that the big lesson here is that grid is just another manifestation - you can use all those other terms - to say that we're moving to a new paradigm for distributed computing. I have this vision, and I'm sure that anybody who's worked in this area probably knew well before I did, but all the work that's going on in the web services resource framework and all that to essentially take grid toolkits and technology and put them into a paradigm that's being developed for the next iteration of the World Wide Web. The Web right now, to me, is a series of static web pages and the capability to do tunneling to be able to complete a very simple transaction. And what gets me excited, though, is imagining years from now, where the Web essentially becomes a backplane of a gigantic, worldwide computer.

GCJ: The kind of computing with no boundaries-type scenario?

Rosenberg: It - the Web becomes the backplane of a computer. You can do anything you want, with unlimited resources. You have unlimited computing power at your fingertips. It makes the possibilities of doing not two-dimensional, but three-dimensional video teleconferences. Right now we're in the early - we're still in the cave, if you will, sitting around the fire. But I can imagine 10, 15 years from now, those little video cams actually produces a conference on the Web, it may become a holographic image, kind of like state-of-the-art visualization that's run on very high performance machines, becomes accessible to any kid.

GCJ: To get to this $24.5 billion spend number, in your opinion, what are the top two or three drivers that are going to push grid to get to that number? What are the reasons that people are going to spend at that level?

Rosenberg: The standards settle down. I think that that may be factor number one and two, is that we have a set of standards that everyone adheres to, that are interoperable. So, two events, the standards mature and the major tools become a little bit easier to use.

close window