Friday, November 24, 2006

Grid today...


This week i read the article posted by gcn with interesting remarks about the strength and weakness related to state of the art of grid computing.

Here, i will summarize the key points on that article and give some final remarks.

The article not intend to be an attack to grid technology, but give important sights about what expectations remain in technological community.

Grid computing, according to the article, is a succeed technology, that has a lot of investments and research. Today, there exist research projects such as SURAgrid Coastal Ocean Observing and Prediction, Globus Medicus; to shown the real possibilities with grid technology.

Globus Toolkit, is the piece of software most widely deployed on grid environments, however looks immature in critical issues such as the installation and management. The installation process is manual and many steps of the procedure is painful.

About, Globus Toolkit the article says: “It is still in its infancy. Automated job submission and resource allocation across a set of machines is still a difficult thing to do,”

Today, there exist software and hardware infrastructure to support the grid technologies, but a long road still remain to cross in order to achieve the Foster and Kesselman's dream.

We envision using idle computational resources, perhaps using new software with grid technology in mind, besides legacy software, we expect to run on the grid platform, too. Licensing problem is an issue in last scenario, because, commercial software is licensed either by user or by machine, therefore, run a commercial software on thousands of cpu machines could be no possible.

Another drawback, is lack of grid support. Globus today is living a similar experience like to Linux were live early in 90s, where only geeks charmed to newcomer technology are the support available at this moment.

However a big difference exist between Linux and Globus. Linux is a technology related to operating systems, concepts are widely studied on computer career, but distribute computing many times is not a core topic of study. Therefore, although there exist many people interested in distributed computing, there exist many technological flavors to deal with that kind of problem. Proprietary, open source, free of licensing among others, populate the internet, but, grid computing begin to gain momentum, and emerges like the most popular distributed technology.

The industry has been doing investment on academy to favor the developments on grid computing. Huge companies as HP, Sun, IBM and Microsoft bid for the new hype of the technology, many times with their own proprietary tools.

You can see Globus Toolkit as the Linux of the grid computing. Many of us, will be testers and will provide ideas, another people around are the engine to keep moving the Globus development.

I hope that this big dream become true, not only with proprietary deployments of Grid Computing, but Open Grid Computing available to do development of the future applications such as games, communication tools, development tools, databases, and so forth.

Finally, i do a couple of questions: Is the grid computing the born of a new paradigm to computation? What new theories has to be developed to support ideas, now applied to grid computing, ideas related to performance, fault-tolerance, adaptivity, uncertainty, and orchestration?

No comments: