[ale] Google innards

[JK]

Greg Freemyer wrote:
> That is amazing.  I bet that has some people thinking.
> 
> Somehow I don't think MS or Yahoo has even thought of running their
> online stuff across that kind of infrastructure.
 >
 > Jim Kinney wrote:
 >> http://news.cnet.com/8301-10784_3-9955184-7.html?tag=nefd.lede


The comments on that article are *so lame* -- there's only
maybe one or two with any substance.

I'm curious why Google is making this stuff public.
Maybe Google feels they're so far ahead that they
won't suffer at all from revealing their voodoo.  Or
maybe they want to sell this tech.  Or maybe (in the spirit
of Not Being Evil) they actually want to inspire some
competition in the realm of massive parallelization?

I read a paper on MapReduce a couple of years ago, and that
is some pretty cool sh*t.  But it's also based on really
ancient ideas: functional languages have had map() and reduce()
for literally 40 years, and some implementations of those have
been parallelized for a long long time -- Parlog is one such.
So they've made a pretty unique pot of chili, using mostly
classic ingredients.

-- JK

-- 
I do not particularly want to go where the money is -
  it usually does not smell nice there. -- A. Stepanov