[kod]

I've been successfully using Spark in production since 0.7, across three or four significantly different projects.

I don't think I could bring myself to ever write another Hadoop job.