On February 25, during a well-produced event at Dogpatch Studios in Potrero Hill neighborhood of San Francisco, Greenplum announced a new Hadoop distribution called Pivotal HD. The following morning, Intel jumped into the Hadoop distribution market with the announcement of its own Hadoop distro, Intel Distribution for Apache Hadoop Software, at an event in downtown San Francisco.
The two moves illustrate, if there was any doubt, that big money can be made in Hadoop. The current competitors in the Hadoop distribution market include no fewer than six vendors: four start-ups, Hortonworks, WANdisco, MapR and Cloudera, and two industry whales, Intel and EMC.
Intel's entrance into the market is particularly telling. The company known for its silicon chips is actually the seventh largest software maker on the planet. So, from Wikibon's perspective, that Intel is entering a software market isn't that surprising. What is surprising is that the company didn't choose instead to partner with an existing Hadoop distribution vendor.
Intel's motivation, however, is not to become the number one Hadoop distribution vendor, in Wikibon's opinion, although surely Intel would be happy to take the top spot and may very well do just that! Instead, Intel is attempting to apply pressure on existing Hadoop distribution vendors and accelerate the market. Intel chips provide the compute power behind Hadoop - regardless of distribution - so the more Hadoop deployments the more chips Intel sells. During the launch event, Intel's Boyd Davis pointed out that for every dollar of software Intel sells, it sells a corresponding $4 in hardware.
That said, Intel has made some significant enhancements to its Hadoop distribution that will appeal to enterprise customers in particular. First, as its Hadoop distribution is embedded in the silicon, so too are the associated security features, such as cell-level security in HBase. This removes the the need for enterprises to apply security features to the database themselves and should result in significant peace-of-mind for CSOs. Intel also says its hardware enhancements significantly increase the performance of Hive, the SQL-like interface commonly used to provide complex query functionality in Hadoop, and the distribution includes tools for automatic cluster tuning.
Intel's Hadoop distribution is generally available now through Intel OEM, ISV and VAR partners and directly by Intel. Pricing is based on an annual subscription and starts at $4,000 per node, putting it within the sweet-spot of Hadoop pricing (though on the high-side of the equation). Initial partners include SAP, which is working with Intel to make its HANA in-memory database work effectively with the new distribution, service providers Infosys and Savvis, and analytics-focused Revolution Analytics and Datameer.
Click here to watch Intel's Boyd Davis discuss the motivation behind Intel's Hadoop distribution live at Strata Conference 2013 inside theCUBE.
Action Item: Ultimately, Intel's entrance into the market validates Hadoop as the real deal and puts pressure on competitors, particularly market leader Cloudera, to close as many enterprise opportunities as possible as fast as possible. For enterprises reluctant to deploy Hadoop due to market immaturity, Intel's entrance should assuage those concerns.
Footnotes: