[Beowulf] A Cluster of Motherboard.

[Chris Samuel]

On Fri, 11 Nov 2005 04:29 am, Robert G. Brown wrote:

> Very interesting indeed.  Of course they DO have the resources and
> economy of scale required to support a machine shop and testbed
> laboratory where they probably try out different vanilla motherboards
> and customize mounting etc as required.  I've also heard that their swap
> out time on a node crash is pretty extraordinary -- minutes.  They have
> hot and cold running techs with spares read to plug right in.

There was a public talk done in Melbourne (at the University of Melbourne) by 
Google's CTO a few years back where he explained where they'd come from (with 
motherboards bolted together with a cheap Lego knockoff) through to what they 
use now.

They use white-boxes for the search side of things with slightly better spec'd 
boxes for the ad-serving.

They can get away with that because they have massive redundancy, the database 
chunks are replicated many times and any query made is sent to multiple 
copies of the database spread across these chunks.

So any failure of any node should, at worst, theoretically, just lead to a 
slowdown of the query.

Caveat Emptor: This is all from memory..

cheers!
Chris
-- 
 Christopher Samuel - (03)9925 4751 - VPAC Deputy Systems Manager
 Victorian Partnership for Advanced Computing http://www.vpac.org/
 Bldg 91, 110 Victoria Street, Carlton South, VIC 3053, Australia

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://www.scyld.com/pipermail/beowulf/attachments/20051111/d70cfa0e/attachment.bin