Orbitz outsources analytics to the cloud

Travel-booking service Orbitz (s oww) has outsourced part of its big data strategy to the cloud  by selecting Kognitio’s Data Warehouse-as-a-Service offering. The decision to move such a critical piece of the analytics stack to Kognitio highlights Orbitz’s commitment to doing big data right, and is further proof that the cloud is an ideal place to do it.

DaaS, as Kognitio calls it this service, is a cloud-hosted version of the company’s WX2 in-memory, columnar row-based analytic database. In-memory analytics technology is particularly hot now, because it takes less time to process information stored in a system’s memory than it does information stored on a hard disk. That means companies can get closer to real-time analysis as data streams in, or far faster results when running less-timely queries. SAP (s sap) and Oracle (s orcl) are also pushing in-memory database appliances.

“Before, it would often take days to get a handle on the data even before we could understand it,” Tony Gray, Orbitz’s director of business intelligence architecture and operations, said in a Kognitio press release. “Now our analysts can almost immediately use the data and drive analytics to find out things that they previously did not know, and use that knowledge to enable the creation of new, leading-edge travel products and solutions.”

Orbitz also uses Hadoop extensively, as detailed in this presentation from Hadoop World 2010. Hadoop serves as a complement to Orbitz’s data warehouse by letting the company store and process even data that might not make it into the Kognitio environment. That means Orbitz can undertake tasks not well-suited to an analytic database, such as machine learning and page download performance, that use various unstructured and not-overtly valuable data types.

That Orbitz chose Kognitio’ DaaS instead of buying either the software or an appliance is a validation for big data in the cloud. Not only does it take a lot of hardware and software licenses to store and process hundreds to thousands of terabytes, but maintaining and powering a big data system costs money, too. Although business models vary, any cloud-based, big-data offering eliminates these issues, while some solutions (but not Kognitio) even mitigate much of the need for data analysts by actually performing the analysis for customers.

That’s why, in some scenarios, big data seems like the killer app for cloud computing. Big data workloads are valuable, but not mission-critical in most instances. That means security concerns around storing data in the cloud might not be prohibitive, and big data systems are big enough and complex enough to justify paying someone else to manage them. Kognitio isn’t the first, nor will it be near the last, to realize this opportunity and capitalize on it.