Dataware for data-driven transformation

Two Wrongs Don't Make a Right

Contributed by

5 min read

I'm sure you will have seen the announcement of the Cloudera and Hortonworks merger by now. I wanted to give readers my point of view on this merger and the opportunities it opens up for MapR.

First, their press release said they aim to "create the world's leading next-generation data platform provider, spanning multi-cloud, on-premises, and the Edge." In fact, this platform exists already - it's called MapR, and we have been delivering it for some time now!

For me, it's a sign that the Hadoop bubble is finally bursting. Don't get me wrong; I see a lot of value in the open source movement, and MapR both supports and contributes to the open source world. It's just that the Hadoop File System was designed for a world of big file batch processing, and modern analytics and applications are focused on small files of human and machine activity, needing real-time capability. And then there is the architectural issue of name nodes in Hadoop that increase the TCO and undermine enterprise-grade SLAs – two reasons why Gartner stated last year that Hadoop was bound to become obsolete because of its complexity.

But there is another dimension to the merger here – the business model is not proving to be sustainable. Two loss-making companies with little to no intellectual property are merging two services-based business models, hoping that cost-cutting will save them. A 3-year roadmap to deliver a 'unity release' that aims to repeat what MapR already delivers is a long-shot bet on pivoting to a product space they have no proven credibility in.

At this point, it's worth noting that the MapR business model, based on productized, patented enterprise software, is ideal for our customers. Our strategy is to embrace our ecosystem and focus on providing the best platform on which everyone can build and deliver new value from their data. We provide a robust and reliable platform that de-risks delivery of complex projects.

The analyst community believes that the 'late majority' is now moving into the world of data lakes and advanced analytics, but they won't move too fast and need the old familiar world brought with them to the new. We are ideally positioned to meet this need in several ways:

  • Most of our customers run critical production workloads on MapR
  • We publish a 5x9s SLA as standard
  • We are <40% TCO than Hadoop
  • We support enterprise APIs, such as NFS, POSIX, ODBC, and SQL, without workarounds
  • We support open source innovations, such as Spark, TensorFlow, Python and R, from the same platform, again without workarounds and the need for extra bolt-on technology

All of the above add up to a true enterprise, integrated, modern data platform. The 'late majority' has a safe place to come to, with world-class engineering, support, and customer references to support this move.

However, one final point - the cloud is not the answer, either. Many commentators state that public cloud vendors are an alternative to Hadoop. Here are a few reasons why I think that argument has flaws, too:

  • The true operational TCO of running in the cloud is still more expensive than on-premises – one MapR customer in Australia used the public cloud to launch and build its new data science application but has now moved it back on-premises for production, as it is cheaper and faster
  • Putting all your data on one cloud service is just another silo in the cloud and means vendor lock-in. Splitting it up between vendors results in more complex technology management and integration challenges

Once again, the answer to these issues is already here – MapR is multi-cloud already and provides inbuilt solutions for data portability as well as data sovereignty.

Many customers have already moved from Hadoop to MapR, and we have a proven approach to support customers and our partners making this transition. Plus, all of our online training is free to access, so that existing Hadoop engineers can learn about the MapR Data Platform and its benefits.

I welcome your feedback and further questions. Existing customers of Cloudera and Hortonworks will now be looking for guidance on the impact of the merger and where they should look to next for help. We are here for them and their partners, tasked with making the right move.


This blog post was published October 22, 2018.
Categories

50,000+ of the smartest have already joined!

Stay ahead of the bleeding edge...get the best of Big Data in your inbox.


Get our latest posts in your inbox

Subscribe Now