On Mon, Apr 5, 2010 at 11:41 AM, Andrew Purtell <[email protected]> wrote:
> The below from Patrick is not uncommon to encounter. > > The "commodity hardware" talk around MR and BigTable is a bit of a joke -- > you can do that if you can afford 1,000s or 10,000s of commodity components > custom assembled. Hadoop+HBase users want to do more with less, obviously. > Colocating computation with storage has its price -- either you horizontally > scale wide or go vertical enough on each node to handle the load you are > throwing at the cluster you can afford. > > I don't think "commodity hardware" is a joke at all, it's just a different definition of commodity. Some people interpret "commodity" as "crap" (desktop class machines). Commodity just means that it's readily available hardware that you can order "off the shelf" from multiple vendors. This is juxtaposed with the old idea of buying $100k+ boxes that were only available from a single vendor (eg big iron IBM or Sun boxes) A <$4k box with dual quad core Nehalem, 24G RAM, and 6x1T drives is very much commodity. Note that I didn't specify any brand/manufacturer there - you can get such a thing from pretty much any vendor (Dell, HP, Supermicro, build yourself, etc), and you can get replacement RAM or drives from newegg that are entirely compatible. This is quite different from the mainframe/big iron world. -Todd
