Skip to main content

Clustered computing

Read this:

http://www.theregister.co.uk/2004/10/15/google_desktop_privacy/

I have my own opinions about privacy, but frankly, there is such a thing as too much privacy. People inherently want to be famous - they have dreams of being movie stars and splashed across pages of magazines and newspapers. They want their 15 minutes of fame...now. The problem with fame and fortune is that you have the media up your nose with cameras. Literally. Someone will take a picture of famous-people's nose hair if they think they can get $15 for the photo. Those famous people are probably embarrassed as I am embarrassed for them as nose hair gets shown on the cover of every magazine. Society couldn't stoop to new lows of sleaze? Forget I ever asked.

However, in the digital world, what people don't realize is that there is power waiting to be tapped and Google is trying to preempt IBM in this regard. (Microsoft is clueless about this and I wouldn't want them trying to mess with something this important - the company is more likely to mess up given its consistent history). Google doesn't honestly care about searching your desktop. What they really want is clustered computing on a grand scale. See, Google runs a huge operation and, if they could outsource their database to your desktop, they could cut the major costs of running their own clustered network and rely on the bandwidth and CPU and memory of all those Windows-based PCs out there. The difficulty is with the details - Google's network is redundant, massively spread out, backed up(?), etc. Someone turns off their Windows PC, what should happen? The best bet is to make the database "float" around the world as the sun travels - people sleep for 8 hours, work for 8 hours, and do whatever for 8 hours...but the time the most PCs are on are those 8 hours of work. So, the database should float on geographical situations and the sun's location in relation to earth. I'm sure small segments would get "lost" here or there but the Google bot would eventually recover it.

The problem is that, to realize this dream, Google has to give people a reason to cluster it. This desktop tool is probably a test, among other things, to see if it is even worth the enormous effort that has to go into building the world's first floating database system. I highly doubt that the desktop tool is all that useful to begin with. Google wants to know how many people will download and where in the world they will download from (they need consistency around the world).

This is a lot of conjecture, but if I were running Google, this would be precisely what I would do to test the waters before pouring effort into developing a floating database. Write a cheesy application for Windows using code we already have and then we wait and see for the next phase. The desktop tool appears to be exactly this. I would honestly like to see if clustered computing on the grand scale both IBM and myself have envisioned is even possible - will people accept it? We will probably find out after Google finds out and eventually does something about it.

Comments