According to Michael Stonebraker and Jeremy Kepner the future of Hadoop is doomed:
|Computational space||Data Management|
|Adopt Hadoop for pilot projects||Adopt Hadoop for pilot projects|
|Scale Hadoop to production use||Scale Hadoop to production use|
|Hit the wall, as the above problems become big issues||Observer an unacceptable performance penalty|
|Morph to something that deals with our issues||Morph to real parallel DBMS|
Let me see if I get this right: you take 2 problem spaces, you generalize these to complete fields, try to use Hadoop, identify the mismatch but still go in production, ignore the solutions built on top of Hadoop/HDFS to address these problem spaces (Apache Hama or Twister) , then conclude by scientific generalization that these problems apply to everyone else, thus Hadoop is dead.
What’s wrong with all these companies using Hadoop for solving their problems? A bunch of stubborn people.
Original title and link: Possible Hadoop Trajectories ( ©myNoSQL)