Well, GridWorld is almost over, and it has been quite an experience. Since I last checked in with my recap of Monday's events, I've had a chance to experience two days worth of presentations, as well as speak with all sorts of smart people spanning the Grid spectrum. In addition, I've had the opportunity to enjoy some good company (and India Pale Ale) after hours, as well as get a feel for this quite spectacular city (and I haven't yet had a chance to check out the national monuments). Granted, a majority of my D.C. experience happened as I inadvertently drove several miles to a Thai restaurant I found online, but it turned out to be a pleasant experience. For starters, I got to see some spectacular architecture and gorgeous views as I left the confines of downtown proper, and I got to feel the vibe of an historic East Coast city (I don't get much of that from my current home base of Las Vegas). Oh, and the kang talay (spicy seafood curry) from Ivy's Place was great, too.

As for what went on in the Washington Convention Center, that was pretty eventful, as well. Tuesday started with Paul Strong discussing how Grid technologies are being used at eBay. For those who don't know, eBay is utilizing Grid on a massive scale. Now, Paul did a Q&A in Monday's issue of GRIDtoday (which you can find either online or in your inbox), so I'll spare a detailed account of what he covered, as a lot is available in that article. However, he did speak about several things that are worth mentioning in this forum. For starters, it's incredible to think about where eBay started and where it is today. Strong noted that the initial site was built from software you can pick up at your local Fry's, had a maximum limit of about 50,000 auction items and didn't even have a search function. Currently, eBay manages a 2 petabyte database, and adds approximately 6.4 million items per day. In addition, it currently has between 15,000 and 16,000 servers in its production environment, which they reprovision every two weeks. As you can imagine, this leads to some serious management issues, he discussed what eBay is doing manage the immense amount of data for which it is responsible and the immense architecture required to manage it. Strong, who also serves as an at-large board member for the OGF, concluded his presentation by discussing why eBay is so supportive of the OGF. That reason, in a nutshell, is that eBay is not an IT tool workshop, nor is it in the business of building management frameworks -- it is an auction site. Hence, widely adoptable standards are imperative if the company is to achieve its goal of replacing its rapidly expanding collection of homegrown software with commercial-off-the-shelf solutions.

After hearing what he had to say, it strikes me as odd that the Grid community is still trying to develop a convincing value proposition for users. After all, Grid has made all the difference in the world for eBay, and is delivering quantifiable results for countless leading companies across a variety of vertical markets. If these often enormous organizations, with the funds to utilize whatever technologies will best suit them, have decided that Grid is the best option, why aren't more companies taking the hint? Perhaps it has something to do with a perception that incorporating Grid will mean an extensive, costly and time-consuming IT overhaul, which brings me to Digipede.

I sat in on Digipede CEO John Powers' presentation, where he detailed how his company is finding success in the SMB market. While Digipede's business plan and use cases have been covered in GRIDtoday more thoroughly than I can do here (I'll point you to any of the company's announcements, as well as this recent article about the company: www.gridtoday.com/grid/803566.html), he made a couple of points that are very noteworthy. His main point, especially when dealing with small businesses, is that it's important to step out of the "Grid world" when pitching solutions to them. One keen observation is that small businesses hate to buy IT solutions -- they simply don't have the time or money to spend running pilots, customizing applications, or negotiating complex sales plans and consulting services. In addition, he noted that very few people in this space have the word "Grid" in their job titles. The bottom line is that the barriers to adopting Grid computing have to be few and far between to get these customers to buy; the closer you can get to offering out-of-the-box functionality, the better.

I had a similar discussion with Appistry vice president of product management and marketing, Sam Charrington, who filled me in on the company's new "Right From the Start" program, which is designed exactly to reduce the entry barriers for SMBs. Basically, companies who qualify will receive three production-use licenses, access to a resource-filled Web portal and dedicated support from Appistry. This company really interests me because they seem to have targeted companies offering Web-based services and running Web-based applications, and are pitching these potential users with the opportunity to be the next Google, Amazon or MySpace. Appistry's EAF seems to fit the bill for companies seeking real-time capable, scalable and fault-tolerant infrastructures, so, similar to what United Devices was able to do in pharma, perhaps we can expect Appistry to carve out a niche in the Web 2.0 space. Like I mentioned before, if you can associate your technology (or a technology) with these major IT users -- Google, in this case -- customers might not be too far behind.

Another highlight of my day on Tuesday was attending the announcements for the GRIDtoday Readers' and Editors' Choice Awards, as well as the IDG Industry Leadership Awards. The event was kicked off by GRIDtoday publisher -- and my boss -- Tom Tabor introducing the awards and commenting on the state of the Grid community, which he has been following since its beginning. Tabor's comments echoed what I believe are the sentiments of many in the community: we've reached an important juncture in terms of both where the technology is headed and how the various stakeholders will interact to advance their collective cause. Regardless of what happens to GridWorld in the future, this year's confluence of the GGF, EGA and Globus communities under one roof conveys a significant message that the technology is here to stay and that the people in the community are passionate, said Tabor. He concluded his remarks by pointing out that the next six months to one year will be very interesting to watch. As for the awards themselves, the list of winners can be found at http://www.gridtoday.com/rca/index.html.

As for the IDG awards for both organizational and individual leadership, I don't believe the winners were a big surprise, but the winners -- IBM and Ian Foster, respectively -- are deserving nonetheless. Certainly, IBM has helped validate Grid technologies with its support of them through the years, and its contributions to GGF and Globus, as well as its various partnerships with Grid middleware vendors and development of its own solutions. As for Foster, what is there to say, really? He literally wrote the book on Grid computing and, if there is such a thing, is something of a rock star in the community. Others in the space have been very influential, as well, but if you're giving an inaugural award for Grid leadership, Foster is pretty much a surefire winner.

Going back to Tabor's comments about what the next year will bring, we received some answers from the OGF, as it announced the dates and locations of its next two events. OGF19 will take place Jan. 29-Feb. 2 in Chapel Hill, N.C., and will focus on group sessions and workshops. Also, because of its locale in one of the leading states in terms of Grid projects, I have to assume there will be a presence by a few of the organizations based there, including the RENCI project. OGF20 will take place May 7-11 in Manchester, England, where it will partner with the U.K. e-Science program and co-locate with EGEE. This event, especially, should turn out to be something really special. OGF, U.K. e-Science and EGEE are three of the most influential organizations and projects around the world, and each has its own impressive roster of contributors, which should lead to a well-attended and very successful conference.

I discussed this decision with Wolfgang Gentzsch on Wednesday, and he cited a couple of key reasons for planning the Manchester event as such, both of which aimed at maximizing attendance -- which has slipped since GGF5 in 2002. In his opinion, though, the drop in attendance wasn't a result of decreased interest in the GGF, but rather due to a significant increase in other Grid projects, organizations and, as a result, conferences and workshops. Heck, the U.K. e-Science All Hands Meeting, Grid 2006 and EGEE'06 all are taking place within the next two weeks in Europe. Even seemingly ubiquitous figures like Wolfgang have to draw the traveling line somewhere. Hence, the combination of these three large -- and often intertwined --  communities should be beneficial to all involved. In addition, Gentzsch pointed out, it is important to plan events as far in advance as possible to make sure would-be attendees have an easier time in planning their trips.

I also spoke with Wolfgang about what he's been up to lately, and was somewhat surprised to learn that he is in the process of scaling back his long list of activities to focus on his work with the German D-Grid initiative, as well an interesting pet project that certainly will make news when it is unveiled. As for the D-Grid, it was recently awarded 5.3 million Euros to build its initial infrastructure, which will consist of several tier-1 centers and a handful of smaller centers throughout Germany. Ideally, Gentzsch said, the project will not only support research, but also serve as a draw for businesses that could make use of the Grid infrastructure. It will be interesting to watch this initiative unfold and join the ranks of functional national grids.

Of the vendors with whom I spoke over the past two days, a common theme among them was data management. Michael Di Stefano, vice president/architect for financial services, at GemStone Systems gave me an overview of what the company has been up to, and highlighted the new features in GemFire 5.0, the company's distributed data management software. You can get a better feel for the new features and the product as a whole by checking out the announcement , but I believe the new version (as well as GemStone's recent partnership with Platform) will go a long way toward helping the company gain some serious traction in the data grid/Enterprise Data Fabric space. I also caught up briefly with Dan Hushon of EMC, who filled me in on what the storage giant is looking to do with the Grid software it acquired from Acxiom acquired from Acxiom in January. While Acxiom provided its Grid solution as a hosted service for customers looking to maximize their consumer data information, EMC will be packaging it for in-house implementations, and believes its advanced capabilities in managing data will be difficult to resist. I'm sure there will be far more details to come as EMC advances its strategy around this software.

Staying in the data grid space, I spoke at length with Cameron Purdy, president and CEO of Tangosol, who had a very unique way of describing the growth of Grid solutions -- including Tangosol's flagship Coherence data grid solution. He described the field of crystallograhpy, where it has been found the crystals on Earth will grow in perfect geometric shapes until they reach a certain point and the effects of gravity alter this course (NASA found that they will continue to grow perfectly in space). Purdy believes that many Grid solutions suffer similar fates, wherein they reach a point where the solutions are ideally suited for their intended purposes, but begin to suffer when customers push vendors into adding additional features into the grid itself. As for Tangosol, Purdy believes his company's crystalline approach was reached in 2001 when the company perfected its peer-to-peer cluster solution. Due to its P2P model, he said, new servers plugged perfectly into the cluster as it grew. However, this "perfect" state was threatened when customers began pushing the company to add a client-server angle to the solution.

With the recently announced Coherence 3.2, Purdy believes Tangosol has tackled this issue without affecting the crystalline structure of the existing P2P mesh. The new version features two new client models -- compute (compute grids) and desktop -- that he describes as "gluing features onto the crystal" as opposed to injecting them into the P2P grid. The compute client, in a nutshell, allows compute grids to pull data across the data grid to solve data starvation issues without being part of the grid. The desktop client, which was created to solve the problem of customers adding PCs (and their laundry lists of issues) to the P2P mesh, utilizes TCP to allow desktops limited access to data in a client-server model without affecting the performance of the grid. In fact, Purdy said that recent testing showed that each node in the grid is theoretically capable of serving 65,000 clients. To illustrate the level of scalability this model provides, a 1,000-server grid would be capable of serving 65,000,000 clients. Coherence 3.2 also features revamped cluster communication capabilities, which Purdy said significantly improve throughput to take maximum advantage of new multi-core processors and advanced 10 GbE and Infiniband interconnects. Pretty interesting stuff ...

Finally, my Wednesday ended with a panel discussion on software licensing in Grid environments, which was moderated by William Fellows of The 451 Group. Throughout its various studies and surveys, The 451 Group has consistently found software licensing concerns to be among the biggest barriers to Grid adoption. The panel included representatives from Grid users (AMD, Intel and Schlumberger), a software provider (SAS), and a company that falls under both categories (Cadence Design Systems). While this clearly is an important topic, the 45-minute time limit of the session wasn't too conducive to an in-depth discussion of possible solutions. The concerns raised by the users were pretty uniform across the board and pretty standard, in general. The bottom line is that they really would like to spend less money on software, and really would like to move away from a per-CPU pricing model if possible. Also of concern is that as users attempt to find ways around growing licensing costs in their Grid environments, such as virtualization, ISVs find new ways to price their software. Ravi Subramaniam of Intel best summed up this particular concern by stating that users end up getting hit coming and going by purchasing virtualization software to alleviate licensing costs, only to be tagged with new, more costly licensing models. As for the ISV point of view, they understand the need to find middle ground on this issue, but contend that it is not necessarily a simple task to develop flexible, user-friendly licensing given the wide range of users and the differing values they place on the software they use.

The discussion started to gain steam toward the end, with discussions about the effectiveness of negotiated enterprise licenses, metered licensing, token-based licenses and even the prospect of having the OGF develop a standardized licensing framework, but, unfortunately, time didn't permit any real fleshing out of these ideas. It's too bad, really, because it isn't too common to have leading ISVs and big-time users discussing this issue in an open forum. I'd be very interested to see this discussion raised again (with more time allotted) in future industry events.

To close, I would just like to remind everyone to read Monday's issue of GRIDtoday, where we will recap the big news coming from the show and I will discuss my experiences from Thursday, as well as offer some final thoughts on GridWorld.