IT organizations spend a lot of time supporting data warehouse applications that are not only expensive to build and support, but they for the most part don't really satisfy the business needs of the people who use them. The fundamental issue is that IT organizations are asked to first model the data stored in a warehouse in a way they think that end users will want to access it. Given the fact that it involves a lot of guesswork on the part of IT, the odds of success are slim to none.
One of things that people really like about Hadoop is that it eliminates the need to do all that because it allows users to invoke analytics applications directly against raw data. One of the latest of these analytic applications arrived this week in the form of Karmasphere 2.0.
According to Rich Guth, vice president of marketing for Karmasphere, Karmasphere is designed to allow IT organizations to get out of the extract, transform and load (ETL) business. Instead of building a massive data warehouse that can be accessed via a report writer that has been programmed to handle a few specific types of queries, Guth says Karmasphere 2.0 utilizes a set of application programming interfaces and Hive data warehouse software built for Hadoop to allow end users to easily query raw data that doesn't need to be modeled first by IT.
The new release also marks the first time that Karmasphere is available as a server-based application. IT organizations can also opt to deploy Karmasphere on a desktop, or in a public cloud service managed by Amazon.
In addition, Karmasphere 2.0 sports a new social interface that is designed to make the application more accessible to a broader number of end users.
There's no doubt that end users still need help from IT to set up Hadoop. But they don't want to have to go to IT every time they want to ask a question. The fact is that most end users aren't even sure what question they want to ask. They just want to be able to easily examine raw data just to see what they might discover.
As with any new technology, there's naturally a lot of skepticism about Hadoop. But once you understand what Hadoop is all about, it becomes apparent that Hadoop is all about getting IT out of the data warehouse building and report-writing business, at least as far as unstructured data is concerned. That alone should make everybody more than happy enough to give Hadoop a try.