Press Releases

by Alan Beck, editor in chief, HPCwire
August 17, 2001

HPCwire: During the 80s, you wrote quite a few articles about computer benchmarking in the old Supercomputing Review, our hardcopy predecessor. Then, you founded Genias Software in the early 90s, and this firm later became Gridware. It's been one year since Sun acquired your company, so this is a good opportunity to talk about the background of that acquisition, its impact on Sun, on Sun's technology, and on Sun's contributions to the Grid. Why did Sun acquire Gridware?

GENTZSCH: Gridware had mainly three valuable assets: Firstly the Grid Engine technology, formerly known as Codine, a distributed resource management software which enables transparent management, administration and usability of networked computers, as one single virtual computer, to deliver compute power as a network service. The second value we brought in, is in our people who designed and developed this software during the last eight years. And the third is the long-term experience in Grid computing which we had gotten in numerous Grid computing projects, like Unicore, Autobench, Medusa, Julius, and Eroppa, since 1995. Most of these projects have been funded by the German Government or the CEC in Brussels, who certainly are very satisfied now to see this technology back in the community, as open source.

HPCwire: How has the acquisition of Gridware gone? Have you been able to integrate into Sun?

GENTZSCH: I was really astonished: the acquisition of our 25-person company and its integration into the 40,000-person company went very smoothly. After 2 months, last September, we delivered the first Sun-branded product, Sun Grid Engine for Solaris. In January this year, we provided Grid Engine for Linux. Just recently, in July, we put all 500,000 lines of code into open source, supported from the open source specialists at Collab.Net. This included a set of courtesy binaries for all major Unix platforms! And now, my team is helping to build Sun's Grid software stack.

HPCwire: What is your conception of the Grid? How do you see the Grid's current state-of-the art?

GENTZSCH: In short, the Grid is a distributed computing architecture for delivering computing and data resources over the Internet, in much the same way that electricity is delivered over the power grid. It is the next logical step in the technology infrastructure which connects distributed computers, storage devices, mobile devices, instruments, sensors, data bases, and software applications, and provides uniform access to the user community for computing, collaboration and communication. Examples of current Grids are the NASA Information Power Grid (IPG); the DoD Distance Computing and Distributed Computing (DisCom 2) Grid; the NSF NCSA National Technology Grid; NetSolve for accessing and sharing mathematical software; Nimrod for campus-wide resource sharing; SETI@Home for searching for extraterrestrial intelligence; the CERN DataGrid processing Petabytes of particle data per year from its Large Hadron Collider experiment; or the APGrid connecting many computer centers in Asia and the Pacific Rim, in the near future.

HPCwire: What are the most interesting aspects of Grid utilization?

GENTZSCH: In general, we can get compute power like electrical power today, which will deliver applications as services, from the wall plug. The list of applications which currently drive the Grid and benefit from Grid developments is long. It started with large-scale simulations in research, engineering, biology, and drug design. Now the benefits are being seen in the sharing of digital content, application service providing, data mining for both research and business applications, collaborative design, remote usage of experimental instruments, remote medical diagnostics, synthesis of large distributed data sets. And the list goes on and on.

HPCwire: From a business perspective, what are the key markets which will benefit most from a Grid product in the near future?

GENTZSCH: I think there is no such product as "The Grid" itself, which you can sell into specific technical or commercial markets. Very much like the Web. You can't sell the Web. To me, the Grid is simply the evolving next generation of the Advanced Web. And like the Web, the Grid will be ubiquitous. It will simply become the basic IT infrastructure for all markets, where it makes sense. What we can and will provide is the appropriate hardware and software stack -- the gridware -- which helps you run your business on top of these grids.

HPCwire: What, then, are the principal differences between the current Web and the new Grid?

GENTZSCH: Internet and Web technology and their usage have improved so much, as expressed in the exponential laws of Moore, Gilder and Metcalfe, that we now see the evolution of the Advanced Web for computing, collaboration and communication. The Advanced Web is available for any kind of business and private purpose, accessible for anybody, at any time, from anywhere. The Grid is simply the buzzword for the next phase of the IT infrastructure. We (somewhat naively) believe that we are now at a similar stage as the electrical power grids: plug in, switch on, and here we go. There are, however, still a couple of years to go until you simply switch on your appliance and use any information service available, over the Internet or just via wireless access.

HPCwire: However, everybody is already talking about the Grid, and there is already Grid software, and some companies are investing quite a lot of money in the concept!

GENTZSCH: Absolutely. We are ready to build such grids today. But there is no commonly agreed standard, there is no common architecture yet. Currently, there is still a lot of manual work involved to set up and run a prototype grid for computing, collaboration, and communication.

HPCwire: This makes it appear that a number of new business opportunities have clearly emerged. How is Sun approaching the Grid business?

GENTZSCH: Since the Grid is still a highly technology-driven software infrastructure, Sun identifies mainly three customer groups: the early adopters, the early majority, and then everybody else. Currently, we are in the early-adopter phase: Customers are highly technology oriented, like our customers in research and education and in some large engineering companies. These Grid customers are our current focus. We are currently building Grids together with these customers.

HPCwire: In what ways are Sun's Grid activities different from those of other vendors?

GENTZSCH: There are significant differences at several levels. First, Sun's origins were in distributed computing. Originally Sun stood for "Stanford University Network," and Scott McNealy's vision, from the very beginning, was "the network is the computer."

Today, you would say that the Grid is the computer. Second, Sun contributed to network computing and grid technologies, for example with Java, Jini, and Jxta, and with Grid Engine and HPC Cluster Tools technology which have become open source projects just recently. Third, and more than any other company I know, Sun has from the very beginning always been a partner of the research community and of our customers. Sun adopts and creates great technologies, but after our own investments into this, we are giving them back to the community.

HPCwire: What about practical Grid-based products?

GENTZSCH: And that's difference number four: We have a whole bunch of software tools ready for the Grid. I have already mentioned Sun Grid Engine and the HPC Cluster Tools, but there is also the iPlanet Grid Access Portal, the Solaris Resource and Bandwidth Manager, the Sun Management Center, and the software from the newly aquired LSC, Inc., for distributed storage management. In addition Sun fully supports and collaborates with all the otheropen grid solutions, so that Sun Grid Engine is already integrated with the major Grid technologies currently developed and deployed in the Grid community, such as Globus, Legion, Punch and Cactus. We are forming these building blocks into one so-called integratable stack ... which is one of Sun's new three big bets for the next IT era, besides Massive Scale and Continuous Real-Time.

HPCwire: How can our readers obtain these grid tools?

GENTZSCH: Many of these are free and in open source from our web site. In addition,Sun is already actively involved in building a large number of grids, Departmental Grids, Campus Grids, Research Grids, and Enterprise Grids. We have created a customer program which is particularly well-suited for this: The Sun Center of Excellence, where Sun is partnering with outstanding customers; and building grids is often a key component in the first phase of such a collaboration.

HPCwire: Could you describe some of Sun's existing grids and grid collaborations?

GENTZSCH: The EPCC Edinburgh Parallel Computing Center is an example for a Sun Center of Excellence in Grid Computing. It is the location of the UK National eScience Center (along with eight regional centers). Basically, the eScience program is to build a UK eScience Grid which interconnects all these distributed computing centers. Edinburgh uses Sun hardware and will evaluate software building blocks like Sun Grid Engine, the iPlanet Portal Server, Sun Management Center, HPC Cluster Tools, and Forte Workshop to build the next generation Grid infrastructure. Through this infrastructure, Edinburgh will deliver compute power to and exchange expertise with its partners in research and industry all over the UK.

Another example is the OSC Ohio Supercomputer Center, which became a Sun Center of Excellence for High Performance Computing, earlier this year. Together with Sun, OSC is building the Grid infrastructure which enables distributed computing, collaboration, and communication with other partners, e.g. Ohio State University, Universities of Akron and Cincinnati, Nationwide Insurance, and Exodus.

One more example is the Technical University of Aachen, Germany, which is a Sun Center of Excellence for Computational Fluid Dynamics. Among other objectives, the Center will be providing remote access to its large Sunsystem (which will grow to over 2 Teraflops) for researchers on the university campus, much like an ASP ApplicationService Provider. Therefore, one of their Grid contributions is the enhancement of Grid Engine toward a Grid Broker, using the software code available in the Grid Engine open source project.

Similarly, Sun is involved in about a hundred ongoing or upcoming Grid activities and partnerships worldwide. For Sun, the Grid is not anything really new, it just is, as predicted many times, the next evolutionary step of the Internet and the Web.

HPCwire: You've described a lot of activities in universities. When and how will the Grid break out into commercial and industrial computing?

GENTZSCH: Grid computing is already implemented in many industrial settings, on a departmental or enterprise level. It may be called differently, up to now. For example Caprion Pharmaceuticals has recently worked with engineers of my group to implement Sun Grid Engine for proteomics. Grids are also coming into the commercial setting through third parties since many ASPs and ISPs are implementing grid-enabled access to applications.

HPCwire: Where do you see Sun's position with respect to the adoption of Linux in the Grid community?

GENTZSCH: In fact, we in Sun collaborate very closely with the Linux community. Sun's strategy is one of interoperability and common tools, such as StarOffice and Sun Grid Engine. That's why you can download Sun Grid Engine for Solaris AND Linux, freely from Sun's website. Or download the source code and compile it for any platform! Today, in the Grid community, there exist hundreds of great research projects around the Grid environment, like Globus, Legion, Punch, Entropia, Unicore, Cactus, Ninf, Nimrod, and many others. It makes sense that researchers start their projects on Linux, because Linux is easily available and widely used in this community. However, Sun is taking care that this software is also ported and optimized for Solaris and integrated with Sun's Grid tools. For example, beginning two or three years ago the Grid community has already begun development of interfaces to Globus, Legion, Punch, and Cactus into Sun Grid Engine, which was then Codine. Development continues today through our open source and collaborations.

HPCwire: Sun has recently highlighted the so-called "Net Effect" - where the Internet is now fundamentally changing the way we do things. And you have indicated that the Grid is really becoming the technology platform for this. What significant changes can we expect as a result of this Net Effect?

GENTZSCH: The Net Effect now forces us to reinvent the network itself: the data centers, the clients, the applications, and the services. Everything is changing. Everything will be seen in the context of Grids - Campus Grids, Enterprise Grids, Research Grids, Entertainment Grids, Community Grids, and many, many more. The network will be service driven, the clients will be light-weight appliances with Internet or wireless access to any kind of resources. Data centers will be extremely safe, reliable, virtually always available, from anywhere. Applications will be part of a wide spectrum of services delivered over the network, such as compute cycles, tools for data processing, accounting and monitoring, with customized consulting, with additional information and communication tools, and with software which allow you to sell or trade your results. Sun's iPlanet software is one example of such service tools operating on these Grids.

HPCwire: So what comes after the Grid?

GENTZSCH: I strongly believe that there will be no disruptive change in the technology or in its usage, for many good reasons. We have so many new technologies on the table that we will be busy enough to efficiently implement them. This takes time. Then, after we have interconnected all kind computing devices through Grid infrastructures, the next step will be to embed and hide these devices into any kind of thing that serves our daily business and private needs, in our houses, cars, airplanes, clothes, the environment, maybe some even in our body.

HPCwire: What must be accomplished to attain this next level?

GENTZSCH: We are still far away from the Grid infrastructure we envision today. We need to agree on one common unifying architecture for the Grids, and on standards and protocols for connecting all the Grid building blocks. Today, hundreds of computer scientists in the Global Grid Forum are concentrating on these most important tasks. We still have to develop much better software for managing, monitoring, billing and trading of any kind of computing services. And we have to make the computing environments absolutely user-friendly, fully secure and reliable.

HPCwire: Thank you, Wolfgang, for this interview!

Wolfgang Gentzsch is Director of Engineering and Grid Computing at Sun Microsystems Inc. He joined Sun 12 months ago, with Sun's acquisition of Gridware, a US/Germany based software company exclusively focussing on distributed computing. He was cofounder of Gridware in 1999, and founder and president of its predecessor Genias Software, in 1990. For over 15 years, he was a professor of math and computer science at the Engineering College in Regensburg, Germany. At the same time, he was a technical consultant for many of the IT companies focussing on vector, parallel and distributed computing.

Source: HPCwire

Back to Top