« October 2006 | Main | December 2006 »
I visited the Center for Computation and Technology at LSU in Baton Rouge on Monday. With Ed Seidel's arrival, and much funding from the state, there is a rapidly growing group of smart and interesting people (e.g., Gabrielle Allen, Thomas Sterling, Tevfik Kosar, Dan Katz, and Jon McLaren) and also a growing scientific infrastructure and collection of strong projects.
I enjoyed reading a recent article by Matei Ripeanu and friends, "Gifting technologies: A BitTorrent case study." They look at a set of six BitTorrent communities with different properties and policies, and compare and contrast various metrics such as degree of freeloading and relative contribution of most frequent uploaders. Arguably some of the conclusions regarding how best to encourage "gifting" are obvious, but I don't think they all are, and there are interesting insights into the relative importance of different factors.
Could it be that some people haven't heard Rolf Harris' stirring antipodean version (another sample and another) of Stairway to Heaven?
I've already written about our celebration of Globus' 10th Birthday at GlobusWORLD in September. But iSGTW has just come out with a nice article on the same topic, so I'll mention it one last time. The article features a startling picture of a (simulated) earthquake in Los Angeles, generated by the Southern California Earthquake Center's TeraShake system. (For more on that application, see another recent article.)
A press release from USC describes the Globus-based MEDICUS system, to be demonstrated at the Radiological Society of North America (RSNA) meeting in Chicago next week. They claim that "doctors [are] finally able to share digital medical images instantly, nationwide, with full patient privacy protection." It is a press release of course, but as I've commented before, it's a pretty neat system. I'll be at RSNA to see it.
Meanwhile, we have two projects underway in the Computation Institute applying the Virtual Data System to medical problems. In one, we are processing 10s of thousands of mammograms and in the other, hundreds of functional MRIs (e.g., see this article). I'll write more about these projects as we get results.
From Dan Atkins, a pointer to a wonderful source of animated information on world development trends. "Making sense of the world by having fun with statistics." You can also see Hans Rosling's talk on this work at TED. I'm not sure which is more amazing: the ways in which modern visualization techniques can be used to bring dry economic data to life, or the misconceptions that many of us (well, myself, certainly) have about the state of our highly dynamic world.
This month's issue of IEEE Computer includes four articles on system-level science: the integration of diverse sources of knowledge about the constituent parts of a complex system with the goal of obtaining an understanding of the system's properties as a whole. This being IEEE Computer, they focus in particular on information technology (IT) issues involved in achieving scientific goals:
[S]ystem-level science integrates not only different disciplines but also, typically, software systems, data, computing resources, and people. System-level science is usually a team pursuit. Data comes from different sources, different groups develop component models, team members provide specialized expertise, and the often substantial computing and data resources required for success are themselves diverse and distributed. Thus, system-level science itself requires the creation of yet another sort of system that may combine large numbers of both physical and human components.
Today's issue of Government Computer News (not the most gripping title for a publication ...) has a long article on Grid. The subtitle is, although proven in academia and research, grid computing struggles to find a place in the enterprise, and the author discusses at some length both where grid has been successful and where it has yet to catch on. It's mostly a fair analysis, and the comments on continued relative difficulty of deployment are right on (although improving, thanks to tools such as Introduce). I'd suggest, though, that one reason for the challenging nature of grid deployments is often sheer ambition. Projects like caBIG and TeraGrid, for example, are complex. But they are achieving things that have never been done before.
I see that a wonderful video blog from France, Bonjour America, has started up again. It is sad to see that M. Vinvin has abandoned his ambition to meet Clint Eastwood. And he now has advertising. But still funny.
I've been spending a lot of time recently talking with economists--of which the University of Chicago has quite a few. We're running a "Disciplinary Deep Dive" (3-D) look at computational economics this quarter, with lectures and discussions on a wide range of relevant topics.
I'm back from the annual Supercomputing (SC) conference in Tampa. As
always there was a lot of cool stuff going on: despite the name, this
is just a great place to go to see innovation in technology and its
applications. A few things that impressed me:
Science Grid This Week (SGTW) is now International Science Grid This Week (iSGTW). This weekly (duh) newsletter features news and stories on grid technology, grid deployments, and scientific applications enabled by grid. It's easy to subscribe. You should also send it stories on your work. There are many smaller projects that don't get the exposure of the big grid deployments, but they are often just as interesting.
Thanks to Beth Cerny Patino (architect and implementor) and Mike Papka (CI senior fellow, and chair of the CI infrastructure committee) we have a new Computation Institute web site. There is much more to be done, but this is a big step forward. We'll be adding more information in the near future on our fellows and staff; their current activities and future plans; and our perspectives on computation and its current and future role in science and society (among other things).
Here's something neat (and details here).
Hadoop, an open source clone of Google FS and MapReduce, can be run on top of Amazon EC2, a hosting service that allows leasing servers on an hourly basis.
As Greg Linden goes on to say:
Developers may now be able to rapidly bring up hundreds of servers, run a massive parallel computation on them using Hadoop's MapReduce implementation, and then shut down all the instances, all with low effort and at low cost. Very cool.
My colleague Tim Freeman points out that you can run those same VMs on your own resources using the Globus Workspace service.
A recent article on "la Computacion Grid" in a Spanish Newspaper, by Borja Sotomayor. He concludes:
La Computación Grid, a pesar de ser una tecnología con bastante madurez, sigue estando en constante evolución. Actualmente existen muchas grids computacionales (como EGEE y TeraGrid), construidas para dar salida principalmente a problemas científicos, pero todavía no existe «La Grid». De la misma manera que internet nació en el ámbito científico para luego llegar al público general, lo mismo puede esperarse, a largo plazo, de la Computación Grid. Cuando la tecnología madure lo suficiente, será posible que cualquier usuario, desde su ordenador personal de casa, pueda enviar complejos trabajos computacionales a «La Grid», como si tuviésemos un supercomputador en el salón de casa.
Borja asked me to write a few words, too, which I did. I didn't realize my Spanish was so fluent ...
Given that it's Sunday, I will write about fundamentalism, a topic that, like the rapture, I find endlessly fascinating. Quoting Wikipedia:
Fundamentalism is a continuing historical phenomenon, characterized by a sense of embattled alienation in the midst of the surrounding culture, even where the culture may be nominally influenced by the adherents' religion. The term can also refer specifically to the belief that one's religious texts are infallible ...
Don't worry, I'm not going to talk about real religion, but rather about fundamentalism in computing. I first encountered this phenomenon in the 1980s, when I hung out with declarative programmers. (Imperative programs are evil! Declarative languages will overcome!) Since then, I've watched with interest as other true paths are progressively revealed and embraced by communities large and small.
Many of the Globus team will be in Tampa, Florida, next week, for the SC'06 conference. Globus technologies and solutions will be discussed and demonstrated in many booths (I think I counted 23 last year) and in many posters and workshops (TeraGrid Institute, GCE'06, and VTDC'06), and a tutorial.
It's hard to capture all of the activity, but here is a partial list, including a schedule of talks to be presented at the Argonne National Laboratory booth.
I am pleased to see an additional six new projects join the dev.globus incubation process, bringing the total to seventeen. The new projects, from Europe and the US, address service development IDEs, security policy management, campus grids, high-level programming, and local resource manager adapters.
Those of you attending the SC'06 conference next week can learn more about these projects, and the dev.globus process, there.
Dan Atkins is the author of the much-discussed "Atkins report" advocating and defining a US national strategy for cyberinfrastructure. Now he runs the National Science Foundation's new Office of Cyberinfrastructure, and in that position has the opportunity to execute on that report's recommendations--modulo the fact that he doesn't have the $1B budget that his report advocated.
Atkins has a new weblog, CI Topics, with pointers to many interesting documents. I like a recent talk he gave to the NSF Education and Human Resources Directorate. In particular, slides 17 onwards talk about the importance of "VOs"--the technologies and processes that enable communities to form and operate efficiently. I think this is just the sort of work required to scale the benefits of cyberinfrastructure to reach millions of researchers and students. Let's hope Dan can make it happen.
You are looking for a development or research project that will contribute to the Globus open source Grid software? Or you have a project that you would like to see done, but you don't have to time to do it yourself? A new wiki provides a clearing house and meeting place where we hope these two classes of people can get together. There aren't many ideas there yet, but the list should grow quickly.
This wiki is not intended as an alternative to the feature request mechanisms associated with dev.globus projects. Use bugzilla for that.
OGSA-DAI (Data Access and Integration) is perhaps less well known and understood than some other Globus components. However, it is one of the most useful and successful. Developed in the UK, it provides uniform Web Services interfaces to diverse data resources. These interfaces allow clients not only to "access" data, but also to query, update, transform, and deliver it. In other words, they let you specify some pretty fancy server-side operations.
In my new role as Director of the Computation Institute, I have got to know some statisticians. I'm learning interesting things:
Ross Anderson's wonderful Security Engineering has recently become available online. The book is full of fascinating anecdotes and deep wisdom. Anyone interested in what computer security is really about (i.e., systems, not algorithms) should read it.
I also like his new article on the economics of information security. The abstract:
The economics of information security has recently become a thriving and fast-moving discipline. As distributed systems are assembled from machines belonging to principals with divergent interests, we find that incentives are becoming as important as technical design in achieving dependability. The new field provides valuable insights not just into ‘security’ topics (such as bugs, spam, phishing, and law enforcement strategy) but into more general areas such as the design of peer-to-peer systems, the optimal balance of effort by programmers and testers, why privacy gets eroded, and the politics of digital rights management.
In other words, the most important components in any security system are the people, and as economics is the study of how and why people make decisions, is very relevant to computer security. Another reason to get University of Chicago economists involved in the Computation Institute ...
Charlie Catlett wrote recently about the GENI program. A quiz: is this ambitious proposal from the networking research community intended to (a) redesign the Internet, (b) get a lot of new money for computer science research, or (c) build testbeds to support innovative networking research?
The answer is "all of the above." Thus, I find GENI interesting for several reasons. First, there's the research agenda, which is to redesign the core Internet protocols to incorporate security, quality of service, and other good things left out by the original designers.
"The conference formerly known as Supercomputing" (now "SC"), takes place in Tampa, Florida, in two weeks time. I'll be there, and one of my goals will be to talk to as many people as possible about their needs for eScience infrastructure and tools.
Two major Globus-related development projects, the National Science Foundation-funded Community-Driven Improvement of Globus Software (CDIGS) and the Department of Energy-funded Center for Enabling Distributed Petascale Science (CEDPS), are both focused on delivering software that meets urgent needs of eScience projects. Thus it's vital that we get to hear about your requirements, experiences, complaints, ambitions. We're also interested in recruiting additional projects into the dev.globus community development process.
If you have ideas you'd like to share or discuss, please contact me so that we can set up a meeting.
A recent discussion at Argonne with my colleague John Bresnahan spurred the two of us to write the enclosed note. In brief, we explain how easy it is to get and install distinct Globus services. These notes are also available on the Web.
1) Globus is Modular
It is a common misconception that the Globus Toolkit is a monolithic entity. This is *not* true. Each Globus service can be installed and used independently. For example, if you just want GridFTP, you can install just GridFTP. *No* additional software will be installed. You get just the software needed for GridFTP, and *only* what you need for GridFTP.
I asked Malcolm Atkinson for his views on this point. He replied as follows: