March 16, 2016 | Written by: Paul DiMarzio
Share this post:
If you’re responsible for analyzing data, you are almost certainly acquainted with Apache Spark. There’s a strong chance you’ve already begun a development project, or at least kicked the tires.
Very often Spark is positioned as a “better Hadoop,” but people rarely write about the aspect of Spark that is poised to revolutionize the way you will analyze operational data going forward. By breaking the tie between analytics library and underlying file system, and by adopting a federated analysis architecture, Spark enables analysis of data in place, on the system of origin, without movement.
Why is this not getting more attention? Probably because for decades we’ve been trained that in order to analyze data, the data must be copied, reformatted and consolidated. It’s all this data movement and manipulation that introduces the latency, cost and risk that can cause insights to perish before you can act on them.
But what if you could perform analysis hand-in-hand with the transactional systems that house your most valuable data, while simultaneously drawing contextual insights from other data sources, without having to go through the pain of data wrangling first?
New announcements from IBM this month will give you the chance to find out for yourself.
This past December, the z Systems analytics team brought Forrester analysts Mike Gualtieri and Richard Fichera to Somers, New York, for a full-day deep dive into IBM’s plans for Spark on the IBM z Systems mainframe. On February 11, 2016, they published their first take in a brief titled “Apache Spark On IBM z Systems Can Reveal Immediate Insights From Transactions”; the title of this blog post reflects the words they used to open that brief. If you’re not a Forrester client, keep an eye on the Real-Time Analytics solution page, or click here for a licensed copy.
All the mainframe Spark goodies that z Systems analytics team relayed to Mike and Richard are being unveiled this month, and the IBM team will be on the show floor at Strata + Hadoop World in San Jose—along with our partners—to talk you through this exciting new mainframe development and answer your questions.
Here’s what you can expect this month:
March 22: A new z/OS Spark product announcement that includes the industry’s only mainframe-resident data abstraction layer designed to make z/OS data available to Spark analysis quickly and conveniently.
March 28: In a one-day event ahead of Strata, held at the IBM Bluemix Garage offices in San Francisco, IBM Business Partner Rocket Software will debut a new engagement model designed to accelerate time-to-value for mainframe Spark projects. It’s not too late to register for this event! Check out Rocket’s Launchpad site for the details.
March 29–31: Strata will be the venue for unveiling a new strategic partnership around Spark and z Systems, as well as a flurry of corporate communications summarizing the value proposition for Spark on z Systems. Both IBM and our new partner will be on the show floor all week to showcase this new tech and discuss how it can help solve your business problems.
I’ve been blogging on the virtues of integrating analytics with transactional systems for several years now (you can read more here). Spark support on z/OS is the latest tool in the z Systems solution portfolio, and it’s destined to open doors to analysis that you never dreamed possible. If you’re coming to Strata, please look for our sessions and come talk to us on the show floor!