By ray@pageonepr.com
via cloudera.com
Published: Mar 11 2010 / 00:29
nugg.ad operates Europe’s largest targeting platform. The company’s core business is to derive targeting recommendations from clicks and surveys. They measure these, store them in log files and later analyze them. Here's the story of how they arrived at using Hadoop for this – and how jobs that required five days to process now require one hour.



Comments
mafr replied ago:
nugg.ad is a relatively small player on the German market. There are several other companies that operate larger systems for exactly this use case.
One company I know of processes terabytes of log data per day. That's not impressive compared to, say, Google, but it's at least an order of magnitude more than what nugg.ad does.
richhutton replied ago:
It is true at nugg.ad that we don’t store massive amounts of data per day, if you compare us with competitors. But we also do not need to store much data per single click, and that is a deliberate design decision. Call this a less-is-more kind of philosophy, if you want.
We can simply log and process more if need be, by adding a new machine to our Hadoop cluster. That requires no other changes to the system.
Voters For This Link (21)
Voters Against This Link (1)