Here are some of the technologies that companies need to look at if they want real-time big data analytics.
The premise of this article is very true. Harnessing big data in motion is critical in many applications. Hadoop's parallel processing of large sets of data is not the answer today. Relatively static data in the Hadoop Distributed File System processed in bulk does not meet the demands of mission critical real time applications in either government or commercial sectors. And the article goes on to indicate there are solutions to this problem.
But the article fails to reference MarkLogic who has been doing this for over a decade. Today, in concert with Hadoop, data can be processed in bulk and streamed, in real time, into MarkLogic. Alternatively, data can be ingested directly into MarkLogic. Since indexes (many) are derived in real time and stored in both memory and on disk, the data is available instantly through enterprise search, business intelligence tools or customer applications.
Applications for big data in motion are diverse and include homeland defense, derivatives trading to assess and manage risk, access to content in support or new digital media products, real time analytics on cargo shipments and a host of customer relationship management applications including data streaming in from the open web.
Buyers should look for technology that is "enterprise grade", low risk to your organization, powerful enough to meet the growing data demands and customer expectations. The ability to read, write, backup and recover big data within a computing architecture that provides real time, high availability should not be underestimated.
Adam Fowler does a great job at describing a very important aspect of Enterprise NoSQL here (ACID Transactions):
Adam also provides some research he collected on why users are choosing to use NoSQL databases (and not). Notice that streaming data and data volumes are two of the top three reasons in support of NoSQL