Mark remarks:

> At a time when many are consolidating their computing power into fewer
>  processors, Google bucks the trend.  (BTW, Google is the best search engine
>  out there IMHO)  There's no doubt that more computers requires more work.
>  Does having multiple computers give better performance, higher up-time and
>  easier scalability than one or two large, fast systems?  Would cheaper
>  e3000's make this an attractive solution or are there limitations built
into
>  MPE/iX that prevent its use in this kind of architecture?

Google works because the problem it solves is highly partitionable. The same
sort of 8000-CPU array that Google has become would also work very well for
weather prediction, nuclear weapons testing, ecological simulations,
searching for patterns in the chaos of radio noise from the galactic plane,
and the like.

If you can design your business database problems and programs in such a
manner as take advantage of a distributed processing environment, then a
Google-like array will work very well.

Most times however for business problems, it's best to put all of your eggs
in one basket.


>  Google Defies Dotcom Downturn
>  http://www.internetweek.com/story/INW20010427S0010

There's nothing wrong with the idea of the internet. Indeed, I'm getting more
enthusiastic about it every day. I personally think that we may be on to
something good here.

And it's true that most dot-coms would have succeeded if it weren't for these
three small stumbling blocks: (i) they didn't have a business plan, (ii) they
didn't have a product, and (iii) they didn't have any customers. Selling ad
space to other people who were trying to sell you ad space in an endless loop
never seemed to be all that bright of an idea to me. But if you ignored these
three problems, for a bright shining bit of time, they were rip-roaring
successes.

Wirt Atmar

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *