There’s a reason a cold shudder runs down the back of any developer when the term “vendor lock-in” rears its head. Nobody wants to be held hostage by their software provider, forced to accommodate time sucking Galapagos-like systems when they could be focusing on, you know, actual development. Having decided that Apache Hadoop technology was running amok in the enterprise logging up just that sort of enterprise angst, Hortonworks has recently brought together Pivotal and IBM to wrestle the data analytics elephant back in line.
Last week, ‘lock-in’ was the buzzword of the day when Hortonworks, IBM and Pivotal announced at Hadoop Summit Europe that each of their Hadoop based platform products are now aligned on a common Open Data Platform (ODP) core of Apache Hadoop 2.6 (inclusive of HDFS, YARN, and MapReduce) and Apache Ambari software.
Commenting on the mutually beneficial three-way, Anjul Bhambri, VP for IBM’s big data and analytics division, noted that, “Different vendors coming up with different versions of Hadoop is not helping its adoption,” however, she was quick to add that the move the trio are in no way “ganging up” on rivals like Cloudera and MapR. In fact, Scott Yara, President of Pivotal has commented that he hopes Cloudera will eventually join the ODP in this top-down development effort.
Unfortunately, this argument doesn’t have everyone convinced. Speaking to Voxxed at the event, Cloudera’s Alex Gutow voiced her doubts at the rationale for the ODP.
Gotow commented; “Honestly, from Cloudera’s perspective, this really hasn’t been an issue that we’ve encountered, so we view this a solution to a problem that doesn’t exist. From our perspective, Hadoop and its ecosystem has been build by the Apache open source community, and Apache has really been this place where developers – with or without the backing of a company – can come together to build the tools that they want to use and they need.”
Adding that Cloudera believe Apache has done “a wonderful job” in this respect, Gutow explained that the company haven’t really seen there be a need for this idea for the one standard, because “Apache is defining that standard – especially if it’s just boiled down to that one core of Hadoop.”
Furthermore, Gutow raised concerns that in uniting behind one version of Hadoop, the three heavyweights may “slow down innovation, versus actually accelerating it through Apache…I think being able to have companies come together outside of that is an easy way to guide it, without giving back to the community.”
Making a Statement for Open Source
Hortonworks aren’t the first company to have drawn the ire of the community when it comes to open source and enterprise. Reactive specialists Typesafe released their own OSS position statement just last week where they touched on the dichotomy they face in “doing what’s best for open source, and giving people a reason to purchase from us.”
In the statement, Typesafe affirm their commitment to giving back to the community writing that, “We wouldn’t be where we are if it weren’t for our open source communities, and so we see our open source products and communities as things that we must build and nurture.” However, for any company couched in open source, there’s inevitably going to be a point where users of any given software find their goals diverge.
Whilst Typesafe are aiming to be “good” collaborators, the company points out that, “We are not doing charity work for you, any more than you are doing charity work for us; we are working together as colleagues because we have shared goals.”
Pivotal and IBM may be on board with the Hortonworks ‘vision,’ but with MapR and Cloudera refusing to climb aboard the bandwagon, vendors will still need to tailor their software for these rival offerings – and judging from the Cloudera rhetoric, this isn’t a situation that will be changing anytime soon. Ultimately, by its very nature, there will always be tension between big enterprise and the wider community of open source committers. Whether the world is ready for an ODP-driven Hadoop remains to be seen, but in refusing to fall in line, Cloudera are sending a strong message on development of Hadoop “by the vendor, for the vendor.’
You can watch our interview with Alex Gutow in full below: