MapReduce Model for Analytics Heads Mainstream

One approach that is gaining new adherents rapidly is called MapReduce, which is a programming model for dealing with large amounts of data that was pioneered by Google. Now companies such as Aster Data are moving to bring MapReduce to the enterprise by closely integrating the new programming model with existing enterprise applications.

With the release of Aster Data 4.0, the company is launching a massively parallel programming (MPP) database that allows chief technologists to tightly couple any Java, C, C++, C#, .NET, Perl or Python application to an implementation of MapReduce running on a database that has been optimized to process data in parallel. Here's a look at the interface:

Aster Data officials argue that in order to take full advantage of the ability of MapReduce to process huge volumes of data in real time, IT organizations will not be able to rely on their existing SQL database implementations. Nor can they tolerate the processing time usually associated with first moving data to and then processing on a traditional data warehouse. Instead, customers can MapReduce-enable almost any application to drive real-time analytics by porting it over to an Aster Data database that supports both MapReduce and SQL. Aster Data, therefore, positions its offering as the first massively parallel data-application server.

Although the MapReduce programming model may provide more raw horsepower than most IT organizations need, it’s becoming clear that the need for real-time analytics of massive amounts of data is expanding well beyond major Internet companies such as Google. In particular, telecommunications, on-line retailers, financial services and the healthcare sector are all prime sectors for MapReduce adoption.

What’s unknown at the moment, however, is just how mainstream database vendors such as Oracle, IBM and Microsoft will respond as MapReduce continues to gain traction.