Just got back from Hadoop Summit 2012. It was an extremely energetic event this year. I was hoping that the HortonWorks people would have more videos from the conference up by now and somethings seems to be holding them up so this will be brief for the while.

There were activities on all fronts:

1. 'Testimonials' from the likes of Sears and Wal-Mart on their experience using Hadoop in the past two years. The video and slides from Sears are available but the Wal-Mart session seems to be MIA still.

2. Updates and discussions on Hadoop MR1, MR2 ("YARN"), Hive 0.09 and 0.10 and more.

3. Research findings from UCB on their attempt to marry in-memory with MapReduce. The result is Spark (cf Hadoop) and Shark (cf Hive). The session was packed and rightly so. The changeover to iterating completely within memory nets up to 40x acceleration. The video and slides are up.

I also had occasion to attend the Hive User Group meeting the day before the Summit. I have heard and read about the community spirit within Apache; it is very encouraging to see in person so many people who are otherwise competitors cooperating and advancing Hive .

Watch for my recommended playlist for the summit when more videos are up.