Brian Christian1 (Zettaset) suggests two roads for adopting Hadoop:
The first, building the capability internally, seems to hold out the promise of flexibility and control for organizations that employ it. While this has sometimes been the case for some large companies, a variety of studies indicate that even among Fortune 500 companies, less than 20 percent that began Hadoop development succeeded in deploying a solution.
The second approach entails working with a big-data, Hadoop-focused third party to develop a bespoke solution. In addition to eliminating the requirement of enormous equipment and human capital investment, this approach also enables organizations, their executives, and IT staff to focus on their core value propositions rather than being forced to become Hadoop specialists.
It would be easy if the decision what be just about CAPEX vs OPEX. Or on-premise vs managed deployments. But there are tons of variables that must be considered when going the Big Data way. Eventually pretty much everyone will do something around Big Data, but those at the forefront still have to figure out many important aspects.
Original title and link: How to Hadoop: Maximizing the value of big data ( ©myNoSQL)