Ian Foster

On Energy Policy

I recently had the pleasure to meet Vito Stagliano, currently Director of Research at the National Commission on Energy Policy, but previously Deputy Assistant Secretary for Policy Analysis at the Department of Energy. This meeting spurred me to buy and read his book, A Policy of Discontent: The Making of a National Energy Strategy.

Energy policy is not a subject that one might expect to be captivating, but I found the book fascinating. Partly this comes from working at a Department of Energy lab, and partly from my growing interest in how socio-economic modeling can yield insights into climate change responses. However, above all, as Vito comments, "energy policy is Shakespearean, with heros and villians, although often not clear which is which." An unlikely statement, perhaps, but true.

Newton's Three Laws of Graduation

For the Sunday humor entry, Newton's Three Laws of Graduation:

• First Law
• Second Law
• Third Law

Michael Turner on the National Laboratories

Our esteemed Chief Scientist has some interesting and inspiring things to say about the US Department of Energy's National Laboratories. Michael Turner led development of the report Connecting Quarks with the Cosmos, which is still a good read, at least for this computer scientist.

The end of Grid computing?

Guy Tel-Zur announces the end of grid computing, based on Google trends data for "grid computing" vs. "virtualization."

The data are fun, but I'm not convinced. Certainly the Google Trends data captures the fact that "virtualization" has replaced "grid computing" as the most popular industry buzzword. But given that industry has used "grid computing" mostly to mean "cluster computing" (e.g., Oracle 10-G, SGE), that doesn't say too much about grid per se.

Measuring adoption and impact is nevertheless an important goal. Thus we have integrated usage reporting mechanisms into our Globus software. We see continued growth in use, as captured by metrics such as service deployments. We're now trying to understand the underlying usage modalities. We believe that many are concerned with "eResearch" functions other than "federating computers"--e.g., on-demand access to computing [on HPC systems and/or EC2], data distribution, service publication and composition, etc. Do these functions count as "grid"? They do according to our article "The Anatomy of the Grid"--and if you look at the goals of projects such as D-Grid.

It would be interesting to see Google Trends data for just "grid." However, that word alone has too many different uses.

Fred Brooks on "computer science"

I recently came across the speech given by Fred Brooks (author of The Mythical Man Month and other wonderful works) on receiving the ACM Allen Newell Award in 1996. In this speech, titled "The Computer Scientist as Toolsmith," he says many interesting things. I'll quote a couple. First, he writes that "computer science" is above all an engineering discipline, concerned with "systems design problems characterized by arbitrary complexity":

Examples are the intricate demands upon operating systems, or knowledge webs, or computer networks. The arbitrariness is inherent—the requirements and constraints spring from a host of independent minds.

These problems scandalize and discourage those who approach them from backgrounds of mathematics and natural science, and for different reasons. Mathematicians are scandalized by the complexity—they like problems which can be simply formulated and readily abstracted, however difficult the solution. The four-color problem is a perfect example.

Physicists or biologists, on the other hand, are scandalized by the arbitrariness. Complexity is no stranger to them. The deeper the physicists dig, the more subtle and complex the structure of the “elementary” particles they find. But they keep digging, in full faith that the natural world is not arbitrary, that there is a unified and consistent underlying law if they can but find it.

No such assurance comforts the computer scientist. Arbitrary complexity is our lot, and here more than anywhere else we need the best minds of our discipline fashioning more powerful attacks on such problems.

It's a useful reminder that "computer science" is not [just] mathematics or physics, and that there are many challenging things to be done in computing that do not involve theorems or physical laws.

Second, he challenges what he saw as the goals of AI research to replace human intelligence:

If indeed our objective is to build computer systems that solve very challenging problems, my thesis is that
                                IA > AI
that is, that intelligence amplifying systems can, at any given level of available systems technology, beat AI systems. That is, a machine and a mind can beat a mind-imitating machine working by itself.

Someday a computer may beat the world champion in chess. When that day comes, I should like to see the world champion equipped with a powerful and suitable IA chess tool, and then play against the AI system. I’ll bet on the IA team.

Amen.

caGrid article

An article on iSGTW describes the National Institutes of Health's cancer biomedical information grids (caBIG) project and its caGrid infrastructure. Not too much new information relative to previous posts, but a good reminder of the nice work that this group is doing. The event that spurred this article is the recent release of caGrid 1.1.

Interestingly, recent articles note that cancer rates continue to decline in the US, but increase worldwide--to and estimated 15 million new cases by 2020. That's a lot of people, emphasizing the importance of this work.

From a technology perspective, caBIG and caGrid are exciting because of the extensive and powerful use they make of the Web Services infrastructure developed over the past several years. In particular, I can't resist pointing out that the entire infrastructure is based on Globus software, and in particular its implementation of the WSRF, WS-Notification, WS-Addressing, security, and related specifications.

RAVI project launched; Charles Bacon new IMP chair

The dev.globus open source grid software community continues to expand with the formation of another Incubator, RAVI: Remote Application Virtualization Infrastructure.RAVI leverages the Introduce system to provide GUI-based tools to guide a user through the process of identifying an application, mapping from strongly typed Web Services operations to application arguments, defining authentication and authorization requirements, and deploying a service onto an execution site. (RAVI, by the way, was the name I suggested humorously for this project originally proposed by Ravi Madduri. He tried to change it to RAVE, but that was already taken.)

This brings the total of active dev.globus projects up to twenty-five. Information on all twenty-five can be found at http://dev.globus.org/wiki/Welcome. (If you're interested in starting a project, please contact incubator-committers [at] globus.org.)

Meanwhile, Jennifer Schopf has completed her term as chair of the dev.globus Incubator Management Project (IMP). During her tenure, the dev.globus community has expanded considerably: we owe her a big thanks for her outstanding and tireless work.

Charles Bacon has kindly agreed to step into the IMP Chair role. We wish him the very best for his tenure.

GridWay wins EGEE prize

A belated note of congratulations to our GridWay colleagues at the Universidad Complutense de Madrid. (A dev.globus incubator.) Their demo entitled "GridWay Interfaces for on-Demand Access to EGEE" won the "Best Demo Prize" at the EGEE'07 Conference. The demo showed the several interfaces provided by GridWay to port applications:

• OGF DRMAA standard support (C, JAVA, Ruby, Perl and Python bindings)
• LRM-like CLI
• From a cluster to the Grid (SGE Transfer Queues)
• GridGateWay (WS-GRAM interface to a whole grid infrastructure), which allows end user to access a meta-scheduling instance using Globus commands

I also note that they just announced a new stable release (5.2.3) of the GridWay metascheduler.

NASA Plans Wireless by 2017

Exciting news from NASA. As Kennedy would have said: "We choose to [install Wi-Fi] in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win, and the others, too."

Nobel Peace Prize and the Grid

The Intergovernmental Panel on Climate Change (IPCC) and Al Gore have been awarded the Nobel Peace Prize.

A perhaps underappreciated aspect of this award is that (as I have noted previously) the climate model simulation data that underpinned the IPCC analysis was made available to the international community via the Earth System Grid, which itself uses (among other things) Globus software. Thus the Grid and Globus communities could (if they were less modest) claim a tiny little bit of credit for that prize.

The next phase of IPCC will require the analysis of far more data than in the current round, as models become yet more sophisticated and more scenarios are run. The next phase of Earth System Grid will feature a more decentralized, federated structure to enable this analysis.