Intel releases its own version of Hadoop

Intel has released its own version of open source software platform Hadoop, which can crunch shedloads of data across large numbers of servers.

Dubbed Hadoop after creator Doug Cutting’s son’s toy elephant, the software was created under an Apache licence. Intel has taken the original code and folded it down the dotted lines, placed part B and connected it to part E and come up with something which is different from the instructions and requires less Allan screws to hold up.

Intel vice president Boyd Davis opened his kimono on the company’s own version of Hadoop, saying that the company had significantly improved the performance of the platform “in certain situations”.   This is an odd thing to say. He could mean that the performance is better than the original when it is switched off.

Intel is mostly known for its tangibles, like hardware, than it is for software, but Davis thinks that Hadoop will mean customers will want more Intel server chips and other data centre hardware.

Davis said that Intel’s glorious five year plan is to drive cloud computing so that it could “affect every human being on this planet in this decade”, for better or for worse.

Wired points out it is another example of tech giants drinking open sauce. Oracle offers hardware to run Hadoop, and earlier in the week, EMC unveiled a tool that runs atop the platform. Paul Perez, vice president and general manager of computing systems at Cisco, has said that his company is working to optimise Intel’s Hadoop distribution for use on Cisco’s UCS servers.

Hadoop has come a long way for a yellow stuffed elephant. It started life as a skunkworks project inside of Yahoo, but none of the crows there ever thought it would fly. It quickly spread to other web operations, including Facebook and Twitter. On the social networking front, it is used to analyse large amounts of data but actually feed data into live software. The original idea was to crunch information using cheap, commodity servers, but when Intel, Cisco and Oracle saw it they wanted it to work on own brand-name hardware and the rest is history.

Chipzilla appears to be seeing Hadoop in much the same way as it does Linux. The chip maker is now one of the leading contributors to the Linux kernel and a key partner in making it happen.

What is different in this case is that Intel is selling software. Davis says that the company is selling its Hadoop distribution and offering technical support for it.

He did not say if he expects the software to net much cash, but it might be that he really wants to see it push more servers.