Grid Computing: Growing Pains
Peter Kelly wrote a post about grid computing that is just great. ”A crisis of identity”. In this post Peter lays out how grid computing is not clearly defined because grid computing is a solution looking for a problem. He actually says, “we’re still trying to figure out what the problem is that we’re trying to solve.” (his emphasis) You should read his post because it’s great. (Did I mention that his post is great?)
What he’s written has got me thinking about the good old days...Some of you may have been around when Microsoft first came out with Windows. If you were geeky enough you would have known that Windows was NOT an operating system. It was an application running on MSDOS that provided a GUI that you could run other applications in. Sure it looked like an OS and was marketed as an OS, but it wasn’t. It has been a migration process to move from Windows the application to Windows the OS. If I remember correctly, which I might not be, Windows XP has been liberated from DOS. (Windows NT was created from scratch a full-blown OS.)
So if you’re wondering what my point is...it’s this - grid platforms look an awful lot like Windows the application. The answers will be found as soon as we figure out what the questions are. And like Windows the application it’s going to take some time, some early adopters, integrators, and a whole lot of work.
Taking a look with my "All Knowing Eye", I suspect that in 10 to 15 years most people will wonder what the big deal was because grid computing will be an integral part of their computing experience. Sort of Google on steriods.
Labels: Concurrency
posted by Kim Greenlee @ 12/16/2005 04:36:00 PM
1 Comments:
Hi Kim,
I'm glad to hear someone else agrees with me ;) You make a good point with the comparison with windows 3.1. People today are still thinking of grid as an application (or "middleware" as it's now called), not as an operating system. Until we start thinking of it like an OS, I don't think we're going to see a properly seamless experience.
The diagram you drew shows clearly that there are layers to a platform, with the operating system at the lowest level, then the grid platform and then the applications. This is the way we should be thinking about grid by abstracting away the underlying details through such a layer.
Unfortunately of course developing such a thing is an extremely complex and time/resource consuming thing to do and it's probably going to take another microsoft or another apple to make it happen. There's a lot of different projects going on in research/academia in this area and hopefully someone will eventually come up with something that meets all the needs and can do to grid computing what GUIs did for desktops.
Post a Comment
<< Home