Features:

TOWARD AN HPC RENAISSANCE: AN INTERVIEW WITH ALAN BLATECKY
by Alan Beck, Editor-in-Chief

HPCwire spoke with Alan Blatecky, an expert in Grid computing, middleware, advanced networking and research program development to gain further insight regarding the recent announcement (http://www.tgc.com/breaking/1738.html) that he will join RENCI as Deputy Director.

(1) You were Executive Director of SDSC for less than two years. What were your feelings about SDSC? What induced you to leave? Will SDSC take a different direction without your guidance?

I was attracted to SDSC because of three reasons; excellent staff, great opportunities at UCSD, and the chance to help transition SDSC to become a national model for cyberinfrastructure and the next generation of high performance computing, storage and networking. Although we’ve had some great successes in migrating SDSC into the new world of cyberinfrastructure and have developed a number of interesting new collaborations, my family remained in North Carolina and I found it difficult to sustain a bi-coastal life style. Thus, when the opportunity to work with Dan Reed and RENCI (which is also based in North Carolina) became an option, it was irresistible.

(2) What attracted you to RENCI? What do you hope to accomplish there?

I was attracted to RENCI for several reasons. I really like and resonant with the focus on interdisciplinary research and applications as this is a hallmark of the next generation of research, science and education. The strength of the research and science communities in North Carolina including Duke, NC State and UNC-Chapel Hill, and their interest in collaboration and the use of advanced technologies is also very attractive. The specific strength of the biology and biomedicine communities is fertile ground for new research projects. And, being located in North Carolina is a great plus as I’ve got roots in North Carolina.

I think RENCI has the opportunity to catalyze innovative research programs and projects bridging an array of disciplines and technologies. The collaborative nature and vision of RENCI ensures that these programs will be vital and will have national and international impact, and I’m looking forward to help make them a reality.

(3) Will your RENCI activities focus primarily on Grid technology, traditional supercomputing, or another approach to computing entirely? Currently, what are your primary areas of technical interest?

I think the best short answer is that while RENCI will certainly explore, leverage, and develop a variety of information technology capabilities (including grid, storage or distributed computing), the real drivers are interdisciplinary applications. While we will develop and deploy technology to solve interdisciplinary research and education needs, technology and infrastructure are tools and not the goals of RENCI.

Although I’m interested in several technical areas including networking, middleware, storage and grid technologies, what I find intriguing is that domains and disciplines can now play, and from my perspective, must play, a formative role in technology development. For example, while RENCI will help enable new significant research programs in biology and bioinformatics, these same disciplines can inform and help develop the next generation of technology and infrastructure.

(4) What will the shape of supercomputing be in five years? How will Grid technology evolve during that period? What are the greatest obstacles to the realization of these visions?

It is tough to predict the shape of supercomputing or Grid technology in five years. Clearly there are a number of companies and projects with plans and solutions in process. However, the hype surrounding grid computing is becoming destructive, as grids have tended to be oversold both in terms of when they will be available, and what they can actually do. As a result, one of the largest obstacles facing realization of the grid vision is the inability to manage expectations.

That said, Grid technology will continue to rapidly evolve over the next decade, but sometimes it is useful to review the key technology drivers that will continue to take place and will enable and guide grid development and deployments. These key technologies are; 1) the doubling of chip power every 18 months, 2) network capacities which have been doubling every 9 months, and 3) the tremendous decreases in cost of storage (disk and tape). These technologies will create new value propositions in unexpected areas, and this in turn will impact grid evolution. So, just as the original Internet morphed into unexpected capabilities (and businesses), so will grid technologies.

Perhaps the most important point to be made however, is that grid development and deployment must be looked at from a long-term perspective. The immaturity and the complexity of the technologies require sustained efforts in research, development and deployment. The immaturity of the field and difficult technical problems (to say nothing of the social challenges) means that research and funding programs must look at 5 and 10-year windows. The complexity of the technologies requires the involvement and integration of a wide range of interdisciplinary teams. These teams include traditional and new scientific domains, as well as domains not normally associated with technology development such as the arts and humanities. These teams will not function just as users, but will be architects and developers.

(5) What created your original interest in HPC? What advice would you give to those who contemplate a career in this area?

My original interest came from the increased personal productivity I experienced from using HPC tools and capabilities and from being able to collaborate with other people. From my perspective, although HPC and Grids in general are terrific technologies, they are simply tools to be used. The real differentiating factor in the end, is expertise and knowledge. Grid technologies significantly enable access and use of distributed resources (computers, instruments, databases) and they also enhance and facilitate collaboration (AG nodes, shared databases, facilities). Together they enable people to work more effectively with one another, and much of science and research will be accomplished at the intersection of these interdisciplinary collaborations.

My advice to those interested in a career in HPC may sound a little odd, as the focus of expertise and the experience base doesn’t have to be purely technical. In fact, because of the nature and complexity of the developing technologies, a “Renaissance” background and experience can be extraordinarily valuable. In fact, since the goal of HPC and grid technologies is to increase knowledge and understanding, this is a province common to all disciplines and domains.

(6) Is there anything else you would like our readers to understand?

We are on the literal edge of establishing a new infrastructure which will affect the way we think, work, and play. While NSF calls this cyberinfrastructure and the Europeans call it e-infrastructure (the popular press call it “Grids”), both recognize the importance of continued research and development, and that the benefits for the R&D extend beyond science and technology. This infrastructure has the potential to increase productivity in science, business and in society in general. And, whenever a technology (like the Internet and now Grids) can create more choices and control to end users (in terms of convenience, time savings, flexibility, or competitiveness), there will be significant growth and adoption. Quite literally, it will usher in a host of new services and businesses, many of which we can’t envision at this point.