Talk about old meets new. At the Strata + Hadoop World summit this morning, IBM Corp. pulled back the curtains on a software bundle aimed at making it more feasible for organizations to run Apache Spark on their System z mainframes. Although it may seem like an unusual combination at first, the company believes the demand is there.

Big iron is often used to handle financial transactions and other fast-moving data that the speedy analytics framework is well-equipped to process. Its in-memory storage model provides the ability to query such information upwards of orders of magnitude faster than disk-based alternatives like MapReduce. According to IBM, the new z/OS Platform for Apache Spark includes a hand-picked assortment of tools that makes it possible to take advantage of the engine’s performance in a wide variety of use cases.

The lineup includes most of value-added addons from the upstream project, which make it possible to analyze real-time data streams, run machine learning algorithms and perform graph analysis to understand the relationship among disparate information. IBM is also throwing Spark SQL into the mix to let analysts  write queries in the same structured syntax they use to interact with traditional databases. The icing on the cake is a custom abstraction layer that enables mainframe applications to make use of the engine without requiring significant modification.

The software has the potential to significantly lower the barrier to adoption and thereby expand the appeal of Spark to many organizations that would not have considered deploying it on their mainframes in the past. Judging by the fact that big iron still accounts for a significant portion of IBM’s total hardware revenues, those customers represent a significant market opportunity, and one with effectively no competition at that.

Image via IBM