Convergence

Web 2.0 and the Future of Pervasive Computing

Saturday, March 25, 2006

Network.com NOT for Most Web 2.0 Apps

Sun Microsystems new grid computing offering, Network.com, was introduced this past week with much fanfare. Jonathan Schwartz boldly proclaimed the service as "the first, publicly accessible instantiation of the future of computing." Clearly, this is a good thing for network-centric computing and there has been a great deal of discussion online regarding the impact of the release. Although I applaud Sun for moving one step closer toward their vision of "The Network is the Computer", the solution really is not applicable to most companies that are developing web-based software applications.
Grid programming is not new. Frameworks like the Globus Toolkit have enabled the creation of massively parallel programs for some time now. The problem with grid computing is that it is not applicable to a large set of applications. Why? Well, first off, most existing applications are not written with a grid-infrastructure in mind. You cannot just dump these applications onto the grid and expect them to speed up. More importantly perhaps, is that for many applications computation is not the biggest bottleneck. Sure there are certain applications in bioinformatics and financials services that require some serious number-crunching, but I would argue that the majority of companies developing web-based software have a far greater demand for top-quality, managed hosting solutions, for their respective multi-threaded applications.

If I were Sun, I would create a "out of the box" hosting service using Niagara machines, their current open source stack, and offer utility computing as a premium add-on. That way, software companies can get the value-added service they need most - managed hosting - and as they scale will have unique the ability to easily add additional processing capability. I mean wouldn't it be great to go to Network.com, quickly configure an OS, database, and App Server in minutes and avoid nasty data center issues? Better yet, you wouldn't you love to have the ability to profile application performance and - as the demand increases for your applications - scale your processing with the touch of a button? I know that our company would love this kind of service.

I hope that the folks at Sun wake up and put together something like this. They already have all the parts and are really missing the boat on a pretty large market.

technorati tags: , , ,

1 Comments:

  • At 5:24 PM, Anonymous miami real estate said…

    I agree. They have something good in their hands right now, and they shouldn't take too long to make the move. Because as we all know, nothing lasts forever. Whatever they may have right now, might not be there anymore tomorrow.

     

Post a Comment

<< Home