I believe a good debate is good for everyone. With multiple points of view and polite but vigorous exchanges, you get a broader perspective and a chance to test claims. This is especially true when the claims come from market research which reflects “commonly held beliefs,” but sometimes compiling opinions is not always a reliable reflection of facts.
Recently, I read a survey that Micro Focus commissioned with technology research firm Vanson Bourne on the state of the nation for mainframe users. Personally, I don’t think it makes pleasant reading, making three arguments why times are bad and getting worse in the world of big iron:
- Mainframe technical debt is on the rise.
- The pool of mainframe talent is drying up.
- The final nail in the coffin: Mainframe applications are hard to modernize for new environments, especially mobile.
Let me take a few minutes here to share my thoughts around mainframe technical debt. It’s probably true that this is increasing, but let’s looks at the bigger picture. After 12 years in the field, six years of which were spent warning that the end was nigh, Microsoft has finally terminated support for Microsoft Windows XP. Yet, as PC World reports, analysts estimate that there may be as many as 30 percent of enterprise desktops still running the outdated operating system. Could there be technical debt behind that unwelcome retention? Compare that to z/OS, where IBM has kept up a reasonable cadence of yearly updates and three-year support. That support period increases to just five years, and now there’s a bi-annual release cycle.
Technical debt isn’t purely a platform issue. I believe it’s bigger than that and it ties in with all the resources (financial and human) involved in IT and R&D, and how that balances with business goals. The mainframe component is far from unique.
Now, let me be clear in sharing that the brain drain claim is nothing new. It’s come up every few years since the first time the mainframe was declared dead back in the 1990s. Back then, when most everyone had switched to client/server and the at the end of the decade, the panic was on to rehire the retired and outsourced armies of mainframe specialists for one last blast, to fix the Y2K problem.
Nearly 15 years later, guess what: Mainframes are still here, and so are the people running them. True, there's a bit of a 30s to 40s generation gap in the z/OS teams I meet, but new hires are coming in and experts from other specialties are being redeployed into the mainframe world by most companies that rely on z/OS. In fact, there's enough fresh blood and institutional memory to keep things healthy.
Yes, young people are joining from college. Two CA Technologies people who are watching this happen are Jim Kokoszynski, the head of our Associate Software Engineer (ASE) program and Dave Danner, who has his own team of new recruits doing very well on customer facing projects.
Jim points out that, once again, the problems the mainframe faces are the same as those across the board. Everyone is looking for talent. To obtain strong candidates, he says, CA Technologies has partnered with several schools; some, but not all, with mainframe specific curricula. They’re producing a steady stream of talented people who are enthusiastic about working with mainframes for good career development reasons. “If the mainframe was easy, anybody could do it,” Jim told me, “hence the candidates we see applying to our ASE Program are unique. They are mature: already technically astute, have strong interpersonal skills and are attracted to the mainframe challenge, its 50 years of proven success, and the opportunity the platform provides for business trends of cloud, mobile, and big data.”
What’s more, Jim says, the ASE program is reinvigorating senior mainframe thinking, as the kids bring in the latest thinking in mobile, cloud, UI, and big data.
Dave says much the same thing, pointing out that CA Technologies hires between 50 and 70 new Associate Software Engineers per year, mostly straight from a CS degree course with little or no explicit mainframe content or experience. “We put them through a rigorous seven-week 'boot camp' to expose them to mainframe skills. They do the majority of coding in Java but are just as proficient with mainframe Assembler, COBOL, REXX, JCL, etc.”
With some extra tutoring and hands-on training, he says, “Today’s generation of young IT professionals can easily tackle mainframe tasks, and are quite good at them.”
Dr. Cameron Seay, Assistant Professor at North Carolina A&T, says that from a broader industry and education perspective, the mainframe route is far from unattractive. NC A&T is part of a group of schools involved in Enterprise Computing Community and IT-oLogy, who, Dr. Seay says, have seen an opportunity and run with it.
“The results have been breathtaking,” he says. “We now have about 15 companies that regularly hire our students. While our focus has been infrastructure, we are moving aggressively into the development space and are partnering with our computer science department to feed more graduates into this space.” Nobody feels like they’re cutting themselves off from the mainstream, he says: quite the opposite. “I have not seen a single student who is properly schooled in mainframe technology who is not absolutely in love with development on the mainframe. This has been an easy sell for our students. Jobs are plentiful and well paying: options are abundant.”
Finally, I’m going to stand firm in saying that you can forget, absolutely, the final point from the survey, that it’s hard to build or modernize mainframe applications. That’s complete nonsense. IBM has pushed the System z platform hard, investing heavily to make it the fastest commercially available processor and the fastest Java execution platform. Since the z9 was introduced, Java performance has exploded five times and it hasn’t finished on that curve, with software and hardware engineers working together to boost performance still further.
At CA Technologies, we’ve been here for years, making good use of the mainframe’s exceptional capabilities. CA Chorus Software Manager and CA Chorus management disciplines are based in Java (utilizing specialty engines) and are entirely Web-integrated.
Customers have been taking their z/OS applications and data onto the Web for years. Web services, XML, and JSON (core technologies for the Internet, cloud, and mobile connectivity) are all available for z/OS and core transactional subsystems, like CICS and IMS. Deploy the Web and application layers directly on Linux on System z and you’ll often see both TCO and performance advantages over relying on the complexity of other hardware and network infrastructures.
The respondents to the Micro Focus survey are entitled to their opinions, of course. But as I hope I’ve shown; if you look in the right places you’ll find that the mainframe is alive, powerful, and very well connected.