5 responses

  1. Shaun Connolly
    March 7, 2013

    Good post. Seems you and I zeroed in on some of the same issues with Woods’ article for Forbes. The quote I took issue with in Woods’ article was “If you are a CIO what do you choose? Open source ideology or products that are made to solve enterprise problems by enterprise companies?”
    Enterprise and open source are NOT mutually exclusive; you point out solid examples of enterprise open source, and I point out a few others.
    In my post entitled “Separating Open Source Signal from Enterprise Hadoop Noise”, I discuss Teradata and Microsoft as Hadoop “use-value participants”, and I specifically cover Microsoft’s efforts within the Apache community (ex. one of their committers is the new new Apache PMC VP of Hadoop). I also cover Fork/Fragment versus Unite/Coalesce approaches to the market. EMC’s approach is valid, but it comes at a cost.Details at: http://hortonworks.com/blog/separating-open-source-signal-from-enterprise-hadoop-noise/
    On the topic of SQL on Hadoop, we definitely see the importance. Rather than abandon Hive or throw it under the bus, we launched the Stinger Initiative aimed at investing in the popular and widely used Apache Hive and working with the smart people in the community to make Hive 100x faster for interactive queries. Needless to say, I believe the community can move the needle in very meaningful ways on this topic.

  2. shaposhnik
    March 14, 2013

    Great post touching on very important questions. To me the ultimate question in all of this is what does make an open source technology a roaring success as a opposed to a mere building block. A thriving community of vendors is definitely helpful, but vendors alone is not enough. Anybody familiar with the history of UNIX, X/Open and “UNIX vendor wars” would surely attest to that. I believe the lesson we should take from Linux is that success is really, in large part,  all about community-driven distributions AKA ways to package open source technology in an accessible (and hopefully friendly!) way. If you ask me what relly made Linux I’d have to say in large part it was Slackware, Debian, Fedora and all the other distros that put a community of users and adopters in full control of their own destiny (as opposed to praying for vendors to ‘get it right’). At Apache Bigtop we believe that this is the crucial bit missing in the Hadoop ecosystem right now. We certainly hope that what we’re building is going to be to Hadoop what Debian was to Linux. We would *love* for all the vendors in the world to build on top of Apache Bigtop just like Cloudera and quite a few other companies do already [https://cwiki.apache.org/BIGTOP/powered-by-bigtop.html] but the message that really needs to be carried across is that there’s now a place where the future of Hadoop ecosystem gets defined in a 100% community driven and vendor independent way. I believe in the media celebration of a all the new and exiting commercial offerings of Hadoop distributions this point tends to get lost.

Leave a Reply




Back to top
mobile desktop