Follow Datanami:
July 27, 2015

IBM Adds Spark to z Systems Mainframe

In a bid to promote “mainframe analytics,” IBM said it would support Apache Spark for Linux on its latest z Systems offering while working with data mining specialists to expand analytic software to leverage additional z Systems processing power.

IBM said Monday (July 27) the “enablement” of Apache Spark is part of an effort to expand its z Systems ecosystem. It plans to integrate Spark later this year with its z/OS. Spark support on IBM mainframes and components would also allow data scientists to leverage a common programming framework regardless of the specific data format.

IBM released its z13 mainframe earlier this year. Ross Mauri, general manager of IBM z Systems, noted in a blog post that the z13’s in-transaction analytics were designed to process transactions and conduct analytics simultaneously. That, the company asserts, allows users to influence business outcomes in real-time, including scenarios such as fraud detection, risk avoidance, blocking improper payments and tailored retail offerings.

Along with supporting Apache Spark on its z Systems mainframe, IBM also said it is working with three analytics partners: Zementis on predictive analytics; Rocket Software on data virtualization; and Elite Analytics on analytics projects running on z Systems.

The collaboration with Zementis would allow data scientists to build models in SPSS, R, Python, SAS and other commercial and open-source data mining tools. They could then be deployed into an operating environment on z/OS. Customers could also use Zementis to make operational models across numerous data sources, including IMS, DB2 for z/OS and VSAM. Zementis also plans to enable the offering for Spark on z Systems, IBM said.

The IBM-Rocket Software collaboration includes enabling its Data Virtualization Service for Spark on z Systems. The service is designed to combine multiple, unrelated data sources into a single data source that can be shared with other applications. Rocket also announced plans to support the R open-source analytic software on z/OS.

The collaboration with Elite Analytics includes providing new services aimed at developing and managing specific predictive analytics projects on z Systems. These will include SPSS or Zementis real-time capabilities on z/OS along with projects developed through the Apache Spark framework on z/OS.

The mainframe analytics push follows an announcement in June that IBM would integrate Spark software into the “core” of its analytics and commerce platforms while offering Apache Spark as a service on its Bluemix cloud application development platform.

Along with advancing Spark’s machine learning capabilities through collaboration with Databricks, the company behind the in-memory analytics framework, IBM also said it would open a Spark Technology Center in San Francisco while committing more than 3,500 developers and researchers to focus on Spark-related projects.

The z Systems partnerships aim to provide a “common framework in Apache Spark for working with data, new software through Zementis and Rocket for conducting analytics and new support services through Elite Analytics,” Mauri said.

Recent items:

Why Hadoop on IBM Power?

IBM Finds the Need for (SQL on Hadoop) Speed

Datanami