The say Elephants like to travel in a herd and that seems to be no different with the Hadoop family. After years of sniffing out their path in the US, three major players have made their way to Europe. At the moment, it is somewhat of a Stampede, Hadoop-based big data platform player MapR’s CEO John Schroeder says. His US-based company has already done US$4m in licencing for its Hadoop-based technologies, and now demand is pulling it in new directions around the world.

“It is all about the pipeline,” Shroeder says. “We are seeing Hadoop adoption and big data adoption in the major metropolitan areas.” Why? Shroeder says like the elephant’s stomping path, Hadoop adoption is steeped in history.

“There is a lot of science to Hadoop’s routes. You generally find it wherever there has been a batch predictive analytics engine. Where companies or organizations were running algo-like clustering algorithms, which have been around for a long time, Hadoop is now providing the ability to run those at scale, with lots of data on a frequent basis. Hadoop has made that efficient and affordable. Analytics algorithms, like clustering and recommendation engines, can run against a petabyte of data and frequently. The more you run those algos the more accurate they get. Then you run them against larger data sets which makes them more timely and accurate because of their scale.”

Hadoop has certainly broadened to become more of a multipurpose platform, moving from batch interactive to real-time programming of APIs. And it is finding ever-new sets of use cases. But elephants in this Hadoop market have a highly ordered and structured social fabric – so don’t expect each player to be offering the same thing. Here are three of the major players – MapR, Cloudera and Horton Works. All announced a European expansion this year, and they all play the game in a different way.

MapR
MapR makes its money from commercial sales of patented big data technology. “We built a much stronger platform to run Hadoop on. We have POSIX-compliant storage layer that includes reliability, performance and opens up Hadoop to APIs,” CEO Schroeder says. “Then we provide our management layer at the top which is also based on our patented technology. So we go to market with a differentiated product which allows us to do things like backup recovery, disaster recovery, performance if you use standard industry benchmarks.”

“Cloudera, for example, brings open source with a proprietary management layer, and Horton Works will support the raw open source. But we sell more software license than anyone else.”

HortonWorks
Hortonworks visited London just before announcing a major win with music site Spotify, considered Europe’s largest user of Hadoop.

Hortonworks offers a 100% open source Hadoop platform, and president Herb Cunitz says it is this that makes its platform attractive. “We also have a heavy focus on partners, building joint products together and taking requirements for what they want into the wider Hadoop community,” Cunitz says.

Its initial use cases were in the research and education sector but now it partners with the likes of SAP, which now resells its Data Platform, for enterprise adoption. “We have about 140 partners. These companies value our open source strategy. They don’t feel held hostage. They don’t want to feel locked in to some components they would need from an independent,” Cunitz says.

Where Horton Works makes its money is from its education and support network. “We want to go and enable Hadoop to function, so that Hortonworks becomes the Hadoop distribution platform that everyone standardizes on,” Cunitz says. “You don’t buy our software, it is all free. We get paid in terms of subscription support, which is about 75 to 80% of our business, and 20 to 25% of it is services such as traditional consulting, helping with implementation or training.” He says more than half of Horton Works’ 250-strong headcount works on engineering or support.

Cloudera – sitting in the middle
Apache Hadoop grew out of Google’s MapReduce and File System papers which was created by Doug Cutting and Mike Cafarella in 2005. Cutting went on to become chief architect at Cloudera, founded in 2008 to bring Hadoop capabilities into the business.

It offers both open source and closed source solutions. Cloudera Enterprise is offered on subscription. It uses Cloudera’s 100% open source distribution of Apache Hadoop and its Cloudera Manager for system management and Navigator for data management, technical support, indemnity and open source advocacy.

When Cloudera COO Kirk Dunn visited London and cleared up a few “misconceptions”. “The biggest misconceptions with Cloudera is that it is a cloud-based product offering. We are an enterprise software company. The technology was invented by Cutting in the Cloud, but we figured out most analytic workloads are safe in data centers today. So we offer a way to consume through cloud but in many cases today we have customers running Cloudera in the data centers,” Dunn says.

Cloudera bases its revenue on support services, training and software.