Big JVMs, bigger Hadoop clusters, Lucene and other stories

Some folks are using LinkedIn's Voldemort on 18G JVMs. In production. Yes, you heard it right.

Then, there's the Facebook Engineering team running 1200 node Hadoop/HDFS clusters to store 12 Petabytes. Phew! I wonder how many times a day this cluster keeps falsely reporting node deaths due to GC pauses.

Awfully nice of LinkedIn to contribute some heavy duty Lucene add-ons back in to the Open source domain. LinkedIn's latest talk on Lucene SDForum.

Another HBase/HDFS talk from FOSDEM 2010 NoSQL - My Life with HBase.

Until next time..