• 03/17/2014
    9:06 AM
  • Rating: 
    0 votes
    Vote up!
    Vote down!

Mainframe Brain Drain Raises Serious Concern

As IT pros with deep mainframe know-how approach retirement age, a major skills gap may await enterprise CIOs that still rely on the decades-old technology.
8 Data Centers For Cloud's Toughest Jobs
8 Data Centers For Cloud's Toughest Jobs
(Click image for larger view and slideshow.)

The sharpest young gun in your IT shop might have a precocious set of in-demand skills, from cloud computing to mobile development to Hadoop and other big data platforms. But ask them to find their way around a mainframe, and they'll likely say: "A what?"

Keith Kohl, director of product management at Syncsort, recently encountered this firsthand when speaking with a group of Hadoop pros. "They understood Hadoop, they understood big data, they understood distributed -- like Linux -- systems," Kohl said in an interview. But there was a notable gap in the group's collective expertise. "It became apparent they didn't even know what a COBOL copybook is, which is absolutely required for mainframe data -- it's one of the most common schemas for mainframe files. They don't even understand how to get the structure of mainframe files and use that to ingest mainframe data into Hadoop [or elsewhere]."

While you might not find the mainframe much in today's startup environments, many longstanding enterprises still rely on the machines, which date back to the 1950s. That's especially true in sectors such as banking and financial services, healthcare, retail, and telecommunications, according to Kohl. The lack of mainframe experience among younger generations of IT pros could become a pressing problem for CIOs in those and other fields in the not-so-distant future.

[For another perspective on how to stay relevant in IT, see Network Engineers: Don't Be The Dinosaur.]

"The people that do run the mainframes are starting to retire," Kohl said. "It's not uncommon for the systems programmers, the mainframe developers [and] architects, to have been in their jobs for over 30 years, 35, even 40 years -- which is getting up there."

The underlying problem with aging mainframe pros: There's no army of youngsters eager to fill their vacant seats. As a result, companies that still depend upon the mainframe might soon find it near impossible to hire the right people necessary to support legacy applications, transaction processing, and other related functions.

Figure 1:
Image credit: Sam Beebe on Flickr.
Image credit: Sam Beebe on Flickr.

"If I'm a bank or a financial institution and every trade or ATM transaction goes through a mainframe system, how do I maintain that? That's a huge risk," Kohl said. "If I'm a health insurance provider and every claim goes through the mainframe, how do I keep those systems up and running?"

In a recent Compuware survey, two-thirds of CIOs said that retiring mainframe talent will hurt their business in the form of increased application risk, lower productivity, and more project overruns, among other potential headaches. Most of the CIOs surveyed (81%) believe the mainframe will remain an important part of their business for at least the next decade.

Kohl pointed to two primary causes of a pending mainframe knowledge gap. The first is simply an image problem: Mainframes aren't cool these days, nor is COBOL or job control language (JCL), two core programming languages underpinning mainframe development. It's much hipper to have the likes of Hadoop on your IT resume. By comparison, mainframe skills don't seem to look as good -- even though they remain relevant to many employers.

The second is a corresponding lack of education and training opportunities. Kohl believes the mainframe is disappearing from university curricula and other training programs. Anyone with an Internet connection can immediately begin taking classes in Python or Java from the likes of Udacity, Udemy, and Codeacademy. Good luck finding a COBOL course on those sites.

Kohl believes the solution will come from a coordinated effort among enterprises, vendors, and universities and other educational organizations to develop the next generation of mainframe skills -- and to promote those skills as sought-after among hiring managers. That effort may have a ways to go: Of the CIOs included in the Compuware survey, 40% said they haven't taken any formal steps to mitigate a potential mainframe brain drain. That's only a modest improvement from 2011, when 46% of IT chiefs in a similar survey said they'd done nothing to plan for a mainframe talent gap. Kohl's optimistic that the industry will eventually move to reinvest in mainframe skills development, but his glass is half full, in part as a matter of necessity.

"The industry, from an IT perspective, has to do it," Kohl said. "All of these really vital, mission-critical systems have to be up and I've got to maintain them."

Could the growing movement toward open source hardware rewrite the rules for computer and networking hardware the way Linux, Apache, and Android have for software? Also in the Open Source Hardware issue of InformationWeek: Mark Hurd explains his "once-in-a-career opportunity" at Oracle.


Does author/publisher know what the subject is?

Your picture is just a black and white photo of a telco punchdown rack.  Technology still in use when dealing with copper phone lines or interfacing to RJ45 connections.  Very little to do with mainfraimes.

COBOL and Mainframes - Keep them alive and kicking!

COBOL and Mainframes are robust and have stood the test of time. The self-documenting nature of COBOL is a facilitating factor to keep these critical systems alive for the next decade.

Software maintenance professionals need to be well rewarded and motivated. There shoud be "joy" in sustenance!

Ramesh Hariharan

CompuSharp Inc

Santa Clara, CA

(408) 850 7160

Just do like my company is doing...

Instead of developing the next generation of mainframers, they're using the ones that have now (who are indeed retiring by the boatload), fill the gap with consultants (of marginal quality I might add) while fully doing away with it and as much as possible, go SaaS. 

Re: Just do like my company is doing...
That is one of the standarrd answers just replace the Mainframe with something, like a gazillion smaller servers, however as you prepare to do that be sure to bring a bug checkbook with you, because it is going to cost like crazy.
Re: Just do like my company is doing...

COBOL also creates some limitations, it is a nice secure environment but it was not written to last for so long and does not enable the creations of new services. The best approach would be a balanced approach, by planning for the future while at the same time keeping the current system alive.

Re: Just do like my company is doing...

I met with General Motors this week, and they described moving all their mainframe apps from an outourced provider to in-house staff in October. And this is in the midst of a big app consolidation effort. GM looks to be keeping mainframe apps around and hiring staff to run them.   

Re: Just do like my company is doing...

I get a similar view from financial institutions, provided with the right API it is possible to move from a mainframe but it's a slow process. These API requires developer skills from both worlds and the firms that have embarked on a long term plan will manage to move onto a modern system that does not have a skills shortage.

some grumbling, but maybe other opportunities to come

At a few "old school" mainframe events in the last few months, our CTO Joe Sturonas has definitely come back wondering where the next generation of mainframers will come from. Could be something where many of the traditional jobs do disappear or morph into infrastructure or cloud "management" roles. I know they've done a solid job of marking the generational woes and not-always-bleak times at Millenial Mainframer

Mainframe is dead and so is COBOL

The COBOL language has been expanded to become impossible to teach or learn.

The IBM mainframe is an expensive proposition for what you get and having a proprietary technology locks a firm in to be at the mercy of smarter competitors.


Re: Mainframe is dead and so is COBOL

No way, COBOL is actually very easy to read, document, follow, debug than many languages that are out there. It reads like simple english. I have written many applications in Cobol and it was never tough to create something that stood the test of time, kept on working.

Response to Mainframe Guys

Well, if companies didn't lay so many of us "Old Timers", Cobol, PLEX, BAL, PLI & Systems Programers off in the last 20 years because they were going to "Downsize", go to "Distributed networks" Migrate to Microsoft servers (Crash daily & 3 finger salute them) there would not be a lack of old farts! We seldom complained about working day and night to keep your systems running, we seldom got angry when the phone rang in the middle of the night when you called us in to work on something. We actually enjoyed our work, we loved keeping our systems running and our customers online with all the systems that these Mainframes ran! Schools stopped teaching Cobol and other mainframe classes because it was trendy to learn Windowz crap! Well, look at where that got businesses now! 

Want a solution? Hire back some of us "Over the hill" folks and we would be happy to support your systems. If not, well good luck.......


EX- IBM Sr. Systems Programmer.

Mainframe VS Youngsters

Very good article.

We lack of a complete knowledge and Support for the whole platform: Z/OS, ISV's, and many critical business aplications running inside the BOX. Is sad to said that Im the younger profesional interacting with the mainframe, Moreover interested in the platform.

I think the manufacturers lack of training and pricing strategies,  and this allowed the distributed systems have won road against Mainframe platforms. 




Mainframe Skills

You don't often hear the words mainframe and Hadoop in the same career conversation. It does seem a MOOC would be helpful here to pass along that inter-generational knowledge.

Homework not done - lots of schools teaching mainframe

In addition to mainframe skills, there's a need for business analytics and mobile development skills that are in short supply.  But there are a lot of schools teaching mainframe courses - over a 1000 worldwide:

There's even a big contest annually for students that test their mainframe capabilities.

Re: Homework not done - lots of schools teaching mainframe

IBM has indeed put some programmatic muscle behind the mainframe -- they've got a good reason to do so. Kohl noted in our interview that IBM's mainframe software business is "alive and well" as additional evidence of the technology's continued relevance. The contest you mentioned, "Master the Mainframe," is definitely worth a look for current/future university students:

That said, I'm not sure it's accurate to suggest that 1,000-plus schools are actively teaching the mainframe -- that many schools may have signed on to IBM's academic initiative at the time (the link you shared is from 2011) but that doesn't mean there's a significant academic investment in the mainframe. (Schools can apply for membership in that program free of charge.) This list is probably more indicative of the US schools that are actually doing something with the program versus just signing up for membership, though I haven't asked IBM for confirmation of that:

Bottom line, I think what Kohl and other are hoping for is a more industry-wide mainframe investment in skills development along the lines of what IBM has tried to do.

the mainframe industry is training smart!

After the economy tanked, many of the soon to retire senior mainframers saw their retirement funds seriously diminished and decided to work for a few more years. This has significantly reduced the severity of this mainframe brain drain retirement issue.

Most mainframe organizations are being smart and are insuring their long term mainframe workforce success by setting up their own internal mainframe training programs to cross train other IT personnel with broad spectrum mainframe skills, as well as provide structured mainframe training for new hires. As the industry well knows, there are very few universities and colleges that can turn out a ready to work mainframer.

These internal mainframe training programs use corporate learning management systems to deliver low-cost mainframe elearning from Interskill (the industry's only mainframe elearning source), setting up mainframe mentoring (also low cost) programs to transfer the knowledge and experience of the senior mainframers before they retire, developing internal classroom training and webinars, plus completing classes from IBM's new Global Training Providers (GTPs) like Avnet, Global Knowledge, Arrow and Learn Quest to provide a multi-modal training program that produces the highest quality "zNextGeneration" mainframers in significant enough numbers to neutralize any mainframe brain drain worries.

I've been involved in global mainframe training for over 20 years and am very impressed to see the industry solving its problems in such a smart way. The mainframe will be in good hands for generations to come.    


I was talking to a group of 30-somethings about the significance of the invention of the COBOL compiler and the "mother" of COBOL -- Grace Hopper.  

They looked at me as if I lost my mind.  

We're doomed.


Job opportunities for me...

I have over 20 years on the mainframe and I can easily see working another 20 years in it. Most of the systems engineers I deal with are in their late 50s and 60s. Having the deep knowledge of the mainframe that I have will be valuable in the coming years as companies look for experienced people (but will they pay me for that experience...probably not).

But, I also have learned Java, C#, and HTML and have started writing plugins at work for Eclipse. The plugins interface with SQLServer, DB2 on the mainframe, and our software change management tool, also on the mainframe . We have also written plugins that can submit jobs on the mainframe to do work. With these plugins, our developers can be in a GUI environment, still coding in COBOL, and never have to log onto the mainframe at all.

Waste of electrons
This is such a beat up - if they want mainframers, then pay them and talent will be attracted to the package. IBM and CA have both been pushing their respective training and development programs in Universities plus making applications that 'simplify' mainframe development and management (aka mainframe 2.0 bullshit). Simple supply and demand will settle this argument - while there are fewer mainframe shops, they do more with less (where have we heard that before), and consolidation amongst the classic mainframe customers continues as they all buy each other. There wil be mainframes and the staff to support them around for decades yet.
Please check your photo - those are phone lines and not a mai...

Please check your photo source - the caption that the photo owner posted was "HACKING THE MAINFRAME Just kidding. Marcus Hecht, Ecotrust Network Administrator, reroutes telephone lines for our new FoodHub coordinators, Megan and Leora."  Those phone lines have nothing do with any computer system.


Point taken about the phone lines in the photo. I apologize for any confusion. Laurie

The Issue isn't a skills shortage. The issue is unlocking th...


By Chris O'Malley

I posted a blog June 2012 that is as valid today as it was nearly two years ago.  When all the other platforms have left corporate data centers, the last computer standing will be the mainframe.  Please see my blog as to why: Far from Extinct, Mainframes Will Rule Entperprise Data Centers in the Future.

The critical challenge for mainframe organizations is to find an easy and fast method of including mainframe transactional data into modern analytics engines.  For many mainframe organizations this means translating and transforming hundreds of millions of rows per day and, in some cases, more than a billion of rows per day.  What complicate this effort is the facts that mainframe data structures that are very foreign to Big Data skilled professionals and that the value of this data is most often greatest in the moment that the transaction occurs.  The critical attributes to any solution that bridges the mainframe with an analytics platform such as Hadoop is abstracting away the complexity of the source mainframe data, easily handling the changes that inevitably occur in the mainframe data formats and transform the data at wire speed to support true, real-time analytics efforts.  

In solving these challenges, the mainframe continues its life as a transactional processing powerhouse with the ability to stream and/or batch fashioned data at wire speeds to analytics engines for true real-time insights. 

Over time, the market tends towards the right tool for the job and companies will invest in the necessary skills required.  The mainframe has proven for 50 years and counting it's value as the right tool for transactional systems.  The data integration work between a mainframe source and a data analytics target is achieved better, cheaper and faster by using the likes of VelociData that provides wire-speed data transformation, data quality, data sorting and data encryption.  The analytics work is best done on either a Big Data platform or within a data warehouse machine.

Brain drain?

I find most of these discussions to be naive and based on fundamental ignorance regarding mainframes.  Most of the talk regarding new technologies treats it as if it were merely a fashion or fad.  It is equally interesting to note how often the tone resembles a kind of "head-shaking" as if those "poor corporations" haven't figured out yet how to get off the mainframe.

 So here's a reality check.  The reason the mainframe is still in such widespread use, is because there is no alternative on the horizon.  The simplistic notions of a few distributed platforms isn't remotely prepared to cover 7x24 availability to process billions of transactions a month with sub-second response times.  The sheer volume of data involved as well as support would render most distributed solutions outrageously expensive and the notable lack of expertise among young people regarding the actual requirements of a data center is truly mindboggling.  [...and please spare me the tired old model of Google].

 The reason universities don't teach the mainframe is because IT professors are ignorant regarding the mainframe.  It has often been an irony that many universities use mainframe systems which aren't even supported by the supposed IT scholars, so while they teach on ***x systems, the university itself runs on a mainframe supported by hired staff.  It's pathetic.

This rhetoric regarding the mainframe is approaching 25 years of talk.  If any of it were remotely true, then I would say ... "put your money where your mouth is", because there hasn't been a single credible replacement of anything except the smallest supposed "mainframe" systems.  Let all the experts come out of the shadows and provide these "new" technology solutions.  Yet, when it comes time to actually deliver, I suspect they'll by silent.

The simple truth is that the young "superstars" couldn't even tell someone what a mainframe configuration consists of, let alone have a worthwhile opinion about how to replace it.

MF talent

There are PLENTY of mainframers, not that old, that are beign cut for offshoring.  Tech more than any other field makes excuses about no talent, and requests more offshore people.  It's all about the money, come on people...   

Re: MF talent

As I was reading this article, I was already saying to myself, "The comments will be full of folks whingeing about there being plenty of mainframers who can't find jobs", and I was right.

The ignorance cuts both ways: too many mainframers haven't learned anything since 1981. They haven't figured out how to use Google, refuse to look at Linux, think TCP/IP is something newfangled that they don't need to pay attention to.

This isn't 1980: the mainframe is just one component in the IT infrastructure, and whether it's better for a given use case or not, it doesn't stand alone. Learn something new. Embrace new technologies. Make yourself more valuable. When it comes time to cut someone, who do you think they're going to pick: the cranky old fogey who refuses to move forward, or the cutting-edge kid who's up on the latest whiz-bang stuff?

Paradoxically, the fact that the mainframe keeps running and running doesn't actually help here: it's all too easy to conclude that something that never causes a crisis must not be very difficult to use, or perhaps not even that useful. That isn't an argument for destabilizing the z environment, mind you! But when there's a networking problem involving an x86-based web front-end interacting with middleware and winding up in DB2 on z/OS, it would behoove the z folks to be part of that solution, able to understand at least some of the other technologiies involved, and not just say "Nope, DB2 is fine, your problem, ".

I've been using mainframes since 1975 or so, and that's my primary job still. But along the way, I've become familiar with Windows, Linux, networking, and a host of other technologies, because they make me better able to do my job. My skills in those other areas certainly aren't at the level of my z skills, but they're strong enough that when someone starts talking about a multi-tier application environment, I can understand and contribute.

Yeah, this is politics to some extent. It's also not being a dead shark: the industry keeps moving forward; so should you.

Re: MF talent

Who said you are the only one learning other skills, besides the MF?  Did I say that people are refusing to learn other things, and thus being let go?  

Re: MF talent

While your observation that many old-timers have not kept up with newer technologies, there are also a fair number that have and are quite diversified.  However, the reverse cannot be said.  The younger "talent" knows virtually nothing about mainframes and the institutions responsible for their education are even more ignorant.

So, it's always interesting that mainframers get accused of being "stuck" in their technologies, while no one ever suggests that Windows or Linux folks learn the mainframe.

Good luck getting the genie back into the bottle

The industry can only blame itself for telling everyone that mainframes are dead. Quite contrary, the mainframe model is alive and well even when it is called "cloud".

Why pick on the mainframe?

One attitude that seems to be prevalent in such discussions is that implicit assumption that somehow the mainframe is deficient as a technology.  Yet, what is this supposed deficiency? 

So the argument invariably shifts to cost, without actually demonstrating why alternate technologies should be cheaper [which they aren't] or how they are more flexible [which they aren't].

Instead we continuously see a picture being painted of the mainframe being some old technology as if it were still some huge beast filled with vacuum tubes.

I'm always fascinated by how someone may bring up the historical image of a mainframe taking up a whole room.  Invariably they don't know what to say when I point out that this is precisely what they are now so proud of when they point to something like the Google data centers.  Yes, we've come full circle.

Bridging the Skills Gap

The current IT skills gap is a growing concern. As discussed in the article, young professionals are entering the workforce simply lacking the traditional skills possessed by experienced mainframe pros reaching retirement. There are two issues you mention in the article that I believe we can help address as an industry -- One, the perception of mainframe and COBOL skills and Two, the lack of education and training opportunities offered to young professionals. As you state, mainframes aren't perceived as 'cool' these days, nor is COBOL coding. However, mainframe systems and COBOL applications are still very critical in supporting the business organization. Did you know that COBOL still powers 70% of all business transaction processing systems, and there are more than COBOL transactions processed every day than Google and YouTube web searches combined!  Many of our 'every-day' activities interact with a Mainframe or COBOL system, oftentimes without us being consciously aware.  Given our dependency on this technology, and the apparent talent gap, who will support these legacy applications?

In order to bridge the gap, academic institutions and organizations must work together to prepare the next generation of workers with the skills necessary to succeed in today's workplace. According to recent research findings from Micro Focus, 85% percent of respondents confirm it's difficult to find staff or new recruits with mainframe application skills. Not only that, but 14% of staff members currently responsible for this task will retire in the next 5 years — up from 11% in 2012. The study also found that the majority of IT leaders (83%) believe it's valuable for students to learn mainframe programming languages and more than 9 out of 10 say these languages should be taught as part of the standard IT curriculum. The reality is that only 27% of universities around the globe adhere to these standards. In order to fill the gap we must teach next generation IT pros the legacy skills necessary to maintain and enhance today's critical business systems.

The current skills gap has actually opened tremendous opportunities for the next generation of IT professionals. For those entering the industry, it is to your advantage that you learn these legacy skills in order to both maintain these mission-critical business systems but also to position yourself for a lasting and rewarding career!

-Ed Airey, Product Marketing Director of COBOL, Micro Focus

Please open your eyes

The author clearly isn't looking beyond the edge of his desk.   As one of the 'old-timers' I have been involved with IBM mainframes for 30 years and I have watched the platform continue to grow and embrace any new technology you want to throw at it.  I have made certain that I keep my skills updated as well.   Any of the dinosaurs you mention are more ostriches with their heads in the sand.  Every discipline will have people who just maintain the status quo but to lump everyone together is an insult.  Also, the author obviously isn't aware of the growing number of new mainframe professionals being groomed, Marist College comes to mind first.   I have to shake my head at the narrow focus of people like the author - just as I do at the people who think virtualization is new....what do you think the mainframe has been doing for 50 years? 

Another Problem

The other problem facing the mainframe world as well as the rest of the IT world is that ubiquitious onager known as a personnel recuiter.  Many of these people are the same people that were unable to cut it in a Computer Science or an Information Systems program in college.  They do however seem to have learned enough IT buzz words before changing their major to history or philosophy to obtain employment as a recuiter, as such positioning themselves to be a oozing carbuncle on the posterior of those who could make it in a CS or IS program.  These recuiters fail to realize that COBOL is pretty much COBOL regardless of platform.  Although I had four courses in COBOL in college, did COBOl exclusively  for about 6 years, I can not now get a job in an IBM COBOL environment because I have never had the opportunity to learn CICS.   Having survived courses in Assembler and Mainframe operating systems, I can't imagine that learning CICS could be any more difficult than either of those two courses.  Yet recuiters will not even present me to a potential emoployer for this reason.  I think that what is needed is a law similar to the real estate law that requires that any offer to buy be presented to the seller, which requires that if a candidate can demonstrate training or experience in the primary skillset specified on a job ad,  that the recuiter be required by law to present the candidates resume to the hiring managers for consideration.  Too many well trained highly experienced people are out of work because some ignoramus in a recuiting firm is looking for cookie cutter candidates.  

The other problem is corporate personnel departments who frequently write position requirements without any direct knowledge of what the technical requirements entail and are incapable of understanding that a candidate who does not have A,B,C,and D on their resume but does have A, B, D and E could function in a given position as well as candidates having A, B, C, and D.  Hiring managers need to wake up.  Their problems as stated in this article are in large measure of their own creation, due to their incompetence and their own lack of professional experience in the technologies that they presume to manage.