Picture of Google's home made server rack from 1999

ic said:
Nope, it's just your eyes :D.

On each shelf [rack position] there are two boards that meet near the middle, underneath each is a thin layer of cork, underneath that is the rack shelf, which does look like it was buckling...
Click to expand...
huh, I can sleep better now :D
 
What they need all the hard drive storage for, Archives?
 
4cer said:
What they need all the hard drive storage for, Archives?
Click to expand...
google keeps a snapshot of most pages over the net, without the photos and other multimedia of course

how do you think the result comes so fast, if it had to connect to 10 servers (10 results per page) each time you do a search in order to show you the extract of the page were the stuff were found, one search would take like 5 minutes
 
I'm sure the millions of and millions of database entries for indexed web pages also take up a bit of space, or am I wrong?

If so, how the hell do they operate?
 
quik said:
I'm sure the millions of and millions of database entries for indexed web pages also take up a bit of space, or am I wrong?

If so, how the hell do they operate?
Click to expand...

They put them in a nice freezing room which has a maxium of -20 :D
 
quik said:
I'm sure the millions of and millions of database entries for indexed web pages also take up a bit of space, or am I wrong?

If so, how the hell do they operate?
Click to expand...

No databases are and can be used except for those that are serving up their ads.
They use their own massive filesystem called ("GoogleFS") on their Linux boxen to spread the files across multiple machines where any node can fail without affecting anything.

Imagine if they had to swap out a broken RAID boxen and restore from last nights backups :D.

Here is a paper written by the Google guys when they started out with their seach engine was still called "backrub" ... some of their ideas are still in Google but probably at a much larger scale.
Secret seems to use lots of machines and spread the information carried.

http://www-db.stanford.edu/~backrub/google.html
 
This may be the reasoning behind buying up "dark fibre" to have yet more locations hooked up to GoogleFS, all working as a team, or automonously. ?
 
PageRank or PR(A) can be calculated using a simple iterative algorithm, and corresponds to the principal eigenvector of the normalized link matrix of the web. Also, a PageRank for 26 million web pages can be computed in a few hours on a medium size workstation. There are many other details which are beyond the scope of this paper.
Click to expand...

And their professors went... "WHAT??!???" :D
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X