In the U.K., exams are over and A-level students are waiting for that Thursday in August when they get their results and know whether they’ll be off to their chosen university. I was chatting with one such student and telling him about mainframes and what they did. That’s when he asked the question, “Why should I think about a job with mainframes?”
For a generation of people who think the “fun” in computing comes from apps on phones, I started to make my case to him for getting a job with mainframes. It’s true the platform is over 50 years old, but that means it’s had 50 years of some of the brightest minds on the planet working on mainframes and developing the hardware and the software ideas that they use. People are still claiming that 70 percent of enterprise data is stored on a mainframe
—which is an impressive figure (although I’ve no idea how it would actually be measured). When it comes to Fortune 500 companies, 71 percent—I guess that means 355 of them—use mainframes for their business. And over 90 percent of the top 100 banks use mainframes. That means successful organizations are using mainframes to keep themselves in business. So getting a job with any one of them would have a future.
The days of green screens and much else that made working on mainframes an arcane skill that was known to some but helped keep others away from the dark arts they performed are gone. Nowadays, of course, it’s very easy to control the mainframe and its subsystems from a multi-colored screen. In fact, displaying monitoring data using browser is fairly commonplace, and those displays are responsive, meaning that they look good on phones and tablets as well as on laptops. The use of things like AngularJS means that the mainframe data displays as impressively as anything else can.
But, you might argue (and my young friend did) that all the exciting computing was being developed by Amazon and Facebook—both non-mainframe using organizations. And it’s true that cloud computing and the ability to crunch massive amounts of data into something usable are associated with those (and other) companies. But, and I did emphasize the “but,” you can do exactly the same stuff on a mainframe. You might picture the look of disbelief he had at this stage.
It started with the IBM z13 launch earlier this year. IBM announced solutions that would enable z13 sites to develop, deploy and manage applications regardless of how they wanted to set up their systems or what types of platform they decided to use. All that corporate data that’s stored on mainframes can now be accessed by cloud and mobile devices. And that’s basically what IBM Bluemix is designed to do. So, while the bulk of an organization’s transactions are running inside CICS or IMS, it now becomes possible to make the information available to the Internet of Things that’s out there.
And talking of Bluemix, it allows programming in Go, Java, Node.js, PHP, Python, Ruby on Rails and Ruby Sinatra. And if my young friend doesn’t learn any of those, other languages can be supported through the use of buildpacks. Then the applications he goes on to write can be deployed using Cloud Foundry apps, OpenStack VMs and Docker containers. That all sounds like cutting edge technology to me.
When it comes to Big Data, IBM is right there. And the fact that it’s using Apache open software makes it quicker and easier for new people coming to mainframes to get to grips with what’s going on. The file system is Hadoop Distributed File System (HDFS). The data store is HBase, which is written in Java. It’s a column-oriented database management system that runs on top of HDFS. HBase applications are written in Java. The runtime is MapReduce, a programming model for processing large data sets with a parallel, distributed algorithm on a cluster. When it comes to workload management your options are ZooKeeper, Oozie, Jaql, Lucerne, HCatalog, Pig and Hive.
IBM has recently started to use Apache Spark. This open-source software can be used to store, process and analyze different kinds of data. Spark will be available as a cloud service for Bluemix. In many ways, Spark is a successor to Hadoop.
In terms of hardware, big data can run on a very large cluster of Linux servers and, of course, Linux on z Systems lets you do this on a mainframe. Android is based on Linux, so working on that part of the mainframe architecture isn’t that different to what you were thinking of doing.
And using an ATM to get out cash means that you’ve been using IMS on a mainframe for years. And so many CICS transactions can be run from a browser that you’ve probably met those without realizing it.
Perhaps it was my enthusiasm for all things mainframe. Maybe it was all those acronyms that seemed to come from the front line of computing. It might have been that I suggested I’d barely scratched the surface of all the other exciting things that mainframes could do. Whatever the cause, this was one student who felt there definitely was a career waiting for him working with mainframes.
Trevor Eddolls is CEO at iTech-Ed Ltd, an IT consultancy. A popular speaker and blogger, he currently chairs the Virtual IMS and Virtual CICS user groups. He’s editorial director for the Arcati Mainframe Yearbook, and has been an IBM Champion every year since 2009.