March 07, 2005
The State of Grid
By all accounts, Grid computing is growing up. While still a creature of the academic and scientific communities in which it has spent a good deal of its early years, Grid's excursions into the commercial enterprise are becoming more frequent and lasting. Take the presentation by Robert Ortega of Wachovia Corp. The following statements from his presentation should speak volumes to the enterprise community and IT professionals who seek to serve it:
"With advancements in Grid technology with respect to transaction processing, we are now leveraging the platform into areas that would be considered more traditional transaction processing; applications in the area of making a trade, creating a customer, locating a trade, retrieving market data, etc."
"The Grid is an ideal virtual application server and an ideal service platform. The Grid enables a high degree of computing efficiency, speed, flexibility, resilience and environment management. Deployment of new services has little to no impact on existing running services, yet is able to leverage existing resources."
We heard similar accounts from companies such as Boeing, SAP AG, Mazda Motor Corp. and Bowne & Co. Grid computing technology seems ready to find a meaningful home within the enterprise community. As Ross Mauri, general manager of On Demand Business for IBM, stated in his keynote address, Grids have arrived:
"Grids are in the business press. Not because they're cool technology, not because they're fun to work with, not because of open source and open standards, but because Grids are solving business problems. So, having Grid on the front page and on the CEO's business agenda is important to me and to all of us."
New Kids on the Block
In the wake of this maturation process, there was a lot of buzz about two additions to the Grid community: Univa Corp., headed by Steve Tuecke, and the Globus Consortium, headed by Greg Nawrocki. Both men come from Argonne National Laboratory, a principal incubator of Grid technology.
Univa, co-founded by patriarchs of Grid computing Tuecke, Ian Foster and Carl Kesselman, will provide commercial software, technical support, and professional services for the open source Globus Toolkit and to the Grid community. Tuecke will serve as CEO, Foster as chief open source strategist and Kesselman as chief scientist.
The Globus Consortium is a new industry group dedicated to the commercial advancement of the Globus Toolkit, an open standards building block for enterprise-level grid implementations. Sponsor memebers of the consoritum are HP, IBM, Intel and Sun Microsystems. Nortel Networks and Univa are contributor members. Nawrocki will serve as president.
While some might question the rationale or implications of the creation of a commercial venture such as Univa, Tuecke said in our interview that it was a well thought-out decision whose time had come:
"The other players, such as IBM, Sun, HP, etc., were saying 'You guys have to get in the game. If Grid is going to happen in the commercial enterprise, you have to do something like Univa. We have to have someone we can partner with commercially to push along the Toolkit and Grid, in general.' There were things we needed to do, that we just couldn't do while with Argonne or Globus."
A lot of spadework was done in preparation for the launch of the consortium. "The community seems to recognize this is a big endeavor that demands community cooperation to help everyone out," Nawrocki said. "We put in a lot of up front work dealing with intellectual property issues, business issues, etc."
The consensus of conference participants is that the emergence of these organizations, particularly with the involvement of the folk formerly with Argonne and Globus, means the Grid is growing up and ready to spread beyond academia and the laboratory into the world of commercial business.
So, with the Grid horizon apparently aglow, where does a developer start? I think Tuecke stated it quite well, taking off on a long stream of thought during our interview:
"Just dive in. We are agnostic to any one particular programming environment or model; this is by design, with respect to Web services and Grid. You don't have to be a J2EE head to do interesting stuff with Grid. Python hackers are doing some nice work around twisted framework with WSRF that integrates well with Globus-compliant services. Client-wise, with respect to distributed data management, try wiring together some workflow. Here, again, use whatever you want -- Java technology, Python, whatever."
"One recommendation I'd make is don't focus as much on doing your own services, but, rather, focus on using existing services. One of the places where a lot of people have tripped up over the past few years, especially since we've been driving OGSI and WSRF, is thinking that Grid is about writing your Web services in compliance with WSRF. This is plumbing-level stuff supporting the interesting stuff like distributed data management, discovery and distributed execution management."
"The place to really start with Grid is to grab Globus and use GRAM, GRIDftp, RFT, RLF and others. The interesting higher-level stuff that is what the purpose of Grid is. Once you have your head around that and start spotting the gaps that get introduced -- that those don't cover -- then looking at doing your own services to fill those gaps makes more sense. Pick your favorite application, like music sharing or whatever, and try to Grid-enable it. One guy developed an online jukebox using GRAM. Get creative and have fun with it."
Foster's take: With the Toolkit building heavily on Web services technology and WS-standards, a developer really needs to know Web services. Further, focus not only on delivering functionality but keep your eye on performance. Performance is extremely important.
Foster also offered this observation, predicting opportunities in the commercial Grid arena:
"I think there is a lot of innovation to come in the area of business models underlying how we accomplish certain tasks and rethinking the way one delivers services in the network world."
With respect to the development community, Nawrocki said developers shouldn't go it alone. "It is to their benefit to invest in the Consortium environment. There really are dividends to be derived from working with your competitor(s)," he said. "If you are using the Toolkit, let us know about it, what are you doing, and what you need."
The space that is "Grid" impresses me as one in which opportunity looms large for the developer community -- a lot of ground-floor acreage. The important question for you, the developer, is how to exploit it.
In Tuecke's keynote, he said the beta of Globus Toolkit V4 is due in a couple of weeks. In our interview, Foster said the GA delivery date of April 29 is pretty firm. According to Foster, the major implication of Globus Toolkit V4 is that, whereas in Globus Toolkit V3 Web services were "not so robust," Web services components in Globus Toolkit V4 "are really pretty solid." He added, "The technology is solid enough to build some interesting things. Also, it's the first implementation of WS Resource Framework (WSRF) and Security. You now have the choice to build on Globus services or build your own using some of these other emerging open standards."
Tuecke said, "Today there is still a mix of Web services and non-Web services. Over time, all functionality on the right side migrating to the left side and also for quite a long time we'll continue to support both." It seems that Grid is destined to be a Web-service world.
Here's the primary thought I took away from GlobusWORLD: Sure, standards are important, but keep your eye on the Toolkit and Univa, and let the standards take care of themselves. There are too many standard bodies to keep track of, anyway. Stay open, push interoperability and you should be OK.
About Hal Hensley
Hal W. Hensley is a writer, IT consultant and principal with Hensley & Associates, L.C., in San Antonio, Texas. He caught a wave in the IT world of the mid-'70s and has been riding along ever since. Comments, suggestions, additions and corrections are welcome. He can be reached at firstname.lastname@example.org.
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?