Re: Is there a mainframe skills shortage?

"Frank Swarbrick" <Frank.Swarbrick@xxxxxxxxxxxxxx> wrote in message


This article really lightened my day... It is raining here (bloody
Autumn...Grrr...) and I finally managed to get a COBOL component working as
a Web Service (after days of frustration), so I decided to take a break from
serious work and respond to the article. Since some of you can't see it,
I've reproduced it here and interspersed my comments... Everything in quotes
is from the article and I've used a few dashes to separate out the article
and the responses. A group of equals signs represents the beginining of a
response...If you want to remove my insightful witty observations and simply
reconstruct the article, just delete everything between a row of equals and
a row of dashes ;-)...

"I've got a lot of respect for much of the market research that Gartner
produces -- but in this case I believe that Gartner is just plain wrong. "
Immediately, we realise this is going to be insightful...:-)

As far as I'm concerned, I've NEVER seen anything from Gartner that looked
even remotely probable, never mind likely...Still, it's nice to see that at
least one of the Faithful is backsliding...:-)

"In a recent research note, "Impact of Generational IT Skill Shift on Legacy
Applications", Gartner suggests that a pending, projected decrease in
mainframe-skilled individuals may be a reason to migrate to other,
"more-modern application platforms".
The logic is that as baby-boomer mainframe coders and administrators leave
the workforce over the next five to seven years, mainframe shops
(particularly the smaller ones) are going to have great difficulty managing
their mainframe environments or maintaining legacy COBOL code. Thus, IT
executives should start planning to go to other platforms. "


While this is certainly a consideration, there are many much better reasons
to migrate away from COBOL... (Interoperability would be a major one in my
book...COBOL does NOT play nicely or easily with other languages. Some
others: COBOL does not sit well on the Network (unless the "network" is
being run from a centralized mainframe), COBOL objects are difficult to
share (unless you wrap them as something else, and always assuming you even
took the trouble to learn OOP in the first place...), COBOL has a major
emphasis on maintaining Source Code and this is error prone and


"However, this Gartner report failed to identify which specific skills were
"at risk." Additionally, it failed to identify that there is a skills
shortage across the entire IT industry and not just in the mainframe market.
Furthermore, this mainframe skills shortage problem is geographical; and all
of the new improvements that IBM is making in mainframe management may
actually reduce the number of people needed to manage mainframes in the
future, as well as reduce the skills needed to manage mainframe
environments. "


That seems pretty fair comment. Looks like its "System Programmers" who are
at risk :-)


Based on interviews with IT executives, university professors, IT
recruiters, and IBM:


I had coffe with my Boss, my tutor, my pimp, and our IBM rep.... :-)


a.. Some mainframe skills are indeed in short supply;
b.. Other skills are readily available (especially Java/Linux skills);
c.. The projected need for an army of mainframe-skilled IT professionals
to replace the existing generation of soon-to-retire mainframers may never
COBOL programming

The term "mainframe skills" needs to be better defined. In my research, I
found that IT managers, recruiters, and university professors have generally
separated "mainframe skills" into four groups:

1.. COBOL programmers (applications developers and code maintainers);
2.. Administrators and managers (with CICS, zOS, and systems management
3.. Operations/planning staffs (business/design consultants, DBAs, and the
like); and,
4.. New applications designers (Java/Linux skill sets).
It is estimated that there are between 150 billion and 200 billion lines of
COBOL code in play in the mainframe marketplace today with several billions
of lines added annually.


But it was Gartner who estimated it.... now it has become an "Everybody
knows" urban factoid. (At least in COBOL mainframe circles...)


Despite rumors of its forthcoming demise, COBOL development is not going
away anytime soon.


Depends what you mean by "soon". I stated in 1996 that I believed there
would be negligible COBOL development going on by 2015. I stand by it. Seven
years (and a bit :-). Not really encouraging if you are planning a new
career (and if I'm right, of course...) :-)


Still, given the huge base of COBOL code in the market today, IT executives
who run mainframe shops should be very concerned about maintaining COBOL
skills over time. But, is there a critical COBOL skills shortage in the
world today? Will there be a critical COBOL skills shortage in the
foreseeable future?


Insert the word "alleged" between "the" and "huge" in the first sentence...
Even if it ISN'T huge, people who run shops where it is used should be
concerned. I notice the author says "mainframe" rather than COBOL, as if the
two are inseparable...they are not.


As I interviewed IT executives, IT recruiters, and university professors,
the following picture developed: IT executives who are able to outsource
their COBOL development and maintenance claim that there is no COBOL
resource shortage. COBOL skills are "easy to find" in India and elsewhere.
IT executives who cannot take advantage of outsourcing due to security or
legislative restrictions are forced to rely on domestic COBOL programmers
who are in comparatively short supply. These programmers usually make
themselves available on a contract basis and usually at premium prices.


However, the fact that COBOL skills are easily obtainable from outsourcing
is cold comfort to those programmers who put all their eggs in the COBOL
basket and do NOT live in India...

It is interesting to me that the general perception by many in the West is
one of thousands of Indians beavering away in COBOL, as if that was all they
can handle. Such is very much not the case. Indian and Pakstani programmers
are at the cutting edge in Java, C#, C++, OO, Web design and services,
search engine optimization, and functional programming. Many individuals and
companies in India are realising that COBOL is a dead horse and will ride it
for the money, as long as there is a market, but they are not neglecting
their futures either. There are some very bright and capable people
emerging from this part of the world and they are not likely to be satisfied
with COBOL programming for decades, as some of us were (or had to be, given
no alternative...)...

Over the last few days I have been reading white papers and articles about
web services and SOA. Some of these articles are written by people who can
hardly speak English (let alone write it), yet their knowledge and
understanding shines through. Seeing India as simply a place to obtain cheap
COBOL skills is way off the beam. It is not only COBOL jobs that will end up
there. (And all this before we even begin to consider China... :-))


"There is a perception in the United States and the European Union that
COBOL is a dying programming language (in fact, one professor told me that
there is an outright bias against COBOL at some universities). As a result,
the current generation of U.S. and E.U. object-oriented programmers want
little to do with COBOL. And further research showed that few U.S./E.U
universities still offer COBOL courses."


Hardly surprising, given that the COBOL establishment wanted little or
nothing to do with object orientation. Given that the world has voted with
its feet and decided Procedural programming is not the best paradigmn, it is
understandable that they are not interested in COBOL. I don't believe this
is a "bias against COBOL"; rather, it is simple pragmatism. Are the
engineers at Ford and General Motors biased against horses? No, they have
just moved on. (Some of them probably own and enjoy riding horses, too)


"Some enterprises face an additional hurdle ? a requirement to "own" their
COBOL talent; in other words, a requirement to directly employ COBOL
programmers. For instance, several government organizations require that
their computer systems personnel be full-time, salaried employees, some for
security reasons and others to limit expenditures on contract labor. There
is clearly a shortage of COBOL talent in the U.S. and the E.U. ? and having
to find permanent, full-time COBOL help presents a real challenge. For
enterprises with these special requirements, finding and keeping COBOL
talent can be expensive."


If they haven't realized by now that maintaining salaried COBOL people long
term is stupid, then they deserve whatever they get. The security is a myth
(just because someone is salaried doesn't mean they won't spill the beans if
the offer is attractive enough), and the idea that permanent employees are
cheaper than contract resources is patently false, if you consider fixed
overheads, paid holidays, sick leave, pension, insurance, and any other
"perqs" of the job. Yes, a six month contract may cost as much as nine
months full time (or even more), but you only do it for six months, not for
the life of the employee. (Of course if you can persuade some poor soul that
they have a job for life then fire them 3 years down the track, you can
probably win, over contracting your requirement out, but you still have to
sleep nights :-), and how long before the word gets around, and so nobody
will work for you, contract or permanent...?)


"However, some university professors are promoting the message that "COBOL
will make you marketable". Several of these professors mentioned that they
inform their students that COBOL-skilled individuals are able to command
higher salaries than their object-oriented, Java counterparts. Some domestic
U.S. students are buying this message. By comparison, this message is
playing really well at IBM's Shanghai mainframe development lab where COBOL
enrollments are way up and the money chase is on."


While I hope they are right, I seriously doubt it. (Maybe in Shanghai... for
now...). Certainly there is less COBOL skill around than Java, no argument.
But that is just a further incentive to move to Java...


"Mainframe systems administration and management

As I researched skills shortages in the areas of administration and
management, I found hundreds-upon-hundreds of openings posted on employment
sites. These sites show that there is clear demand for mainframe
administrative and management skills. Further, a large number of these
postings often go unfulfilled over a thirty-day period, indicating to me
that enterprises are having trouble filling these positions. "


If there are "hundreds-upon-hundreds" then, obviously, there are not enough
people to fill these jobs, OR, the rates are not high enough to attract
people with the right skills. Why would you just automatically assume the


"However, one IT recruiter told me that "the demand for mainframe skills
pales in comparison to the demand for hardware technicians, help desk staff,
and client/server database administrators ? particularly Oracle and SQL
Server database administrators." The problem of finding individuals with
computer skills is not solely a mainframe problem ? it's a problem across
the entire computing industry. On the day that I visited's Web
site, I found that there were four times as many jobs that needed to be
filled in non-mainframe disciplines. In short, there's a major shortage of
trained IT talent across the industry. "


Yes, that MAY be so, or it MAY be that enough isn't being paid to get the
right people. It also raises the question of how the industry continues to
function WITHOUT these people... Could it be that much of this is simply
"empire building" by middle managers?

It is also true that the skill set needed to run a Help Desk is far removed
from the skill set needed to maintain a z/OS system and keep it running. We
are talking here about diverse skill sets, some of which can only really be
acquired "in the field" (not from a Computing Science course).


"Modern mainframes, operations and planning

Finding mainframe database administrators, business consultants, business
process flow experts, designers, integrators, and testers is difficult. But
this is a cross-industry problem. There are still thousands of jobs posted
for mainframe design, implementation, testing, and communications positions.
This tells me that enterprises are still strategically committed to
mainframes as back-end database servers, transaction engines, and security
hubs. They are not looking to abandon or "re-platform" their mainframes.

Over the past five years, IBM has reinvented the mainframe and endowed it
with new processing capabilities. These capabilities include specialty
processing engines (zIIP and zAAP), as well as the ability to run thousands
of Linux instances on a mainframe platform. This ability to run the Linux
operating environment, and accompanying Java applications, modernizes the
mainframe. It makes it possible to use mainframes to run modernized
(non-COBOL legacy) applications. And, the ability to run these modern
workloads solves a big problem for mainframe buyers because there are plenty
of fresh college graduates available who have been trained on Java/Linux
platforms. If an enterprise purchases a mainframe to run Java/Linux
workloads, that enterprise is likely to experience fewer problems finding
the skills needed to run its mainframes.

Retiring mainframe staff and future managers

I could find no studies that showed how many mainframers are about to reach
retirement age. It is reasonable to expect that the bulk of these
retirements will occur between five and twelve years from now as this is the
timeframe when most of the baby-boomers reach retirement age. These
retirements will happen in a phased manner, and some prospective retirees
will not retire at all, working beyond official retirement age. Add to this
mixture that there are plenty of 35 to 50 year olds involved in managing
mainframe environments today. Not everyone who manages a mainframe is over
60 years old, so there is a "second crop" of mainframe managers currently in
the queue at many enterprises throughout the world.

Also, mainframes are becoming easier and easier to manage. IBM is
simplifying mainframe management and is spending $100 million to give the
mainframe a Windows-oriented, easy-to-use graphical user interface. By doing
this, IBM is not only making it possible for lesser-skilled individuals to
manage mainframes ? the company is also making mainframe management appeal
to our next generation of Windows-born-and-raised managers and
administrators. "


So finally, Big Blue is spending chicken feed to implement a GUI. (Insert
hollow laughter here :-)) Next they'll be implementing DotNET...:-) Hey, we
might even see Web Services running on the mainframe...:-) Perhaps even (30
years late) a decent IDE (and don't tell me about Eclipse being an IBM
product :-))


"The mainframe serves a unique role in the enterprise as a centralized,
secure database hub, as a powerful transaction engine, and as a host of
mission-critical business logic. "


Maybe. But until it can match price with modern Network servers, it will
remain the poor relation. I love it when they refer to mainframes as safe
for "mission critical" applications... like no-one ever got a S0C7 or S0C4
or ever had to pore over a memory dump running down event block chains,
because the fundamental tool set you need for problem solving wasn't
there...ANY application that goes live is "mission critical" to the people
who have to use it. Mainframes have no better a track record with this
regard than Client Server systems. Both have had poorly designed
applications implemented on them with disastrous results, and both have run
well designed systems flawlessly. It is only hardware. Success is determined
by the artist, not the paintbrush (MCM said it here first; I agree
wholeheartedly... :-))


"IT executives know this ? and maybe this is why the mainframe market grew
8% last year. The bottom line: the Gartner suggestions that IT executives
consider "re-platforming" or migrating to other "more modern application
platforms" due to a projected, unsubstantiated shortage in mainframe skills
needs to be re-thought. "

ABOUT THE AUTHOR: Joe Clabby is president and found of Clabby Analytics, an
IT research and analysis firm. He over thrity years of IT experience


I'm wondering if the "8%" came from Gartner... :-) Even if it didn't, I
would be surprised if any credible source shows the mainframe share of the
IT market as even equal to what it was in 1980... (So much for stupid
statistics that really prove nothing...)

Given that Gartner are advising a migration away from COBOL I wouldn't send
to know for whom the bell tolls... This reminds me of certain rodents
departing from vessels that are about to explore the ocean from the "other"
side... :-)


The Gotterdammerung of COBOL is upon us. The temples are being converted
into cinemas and they are showing streamed video from RSS. The High Priests
are being pensioned off or encouraged to early retirement; the truly Holy
amongst them who understand the Nature of Things have already expanded their
skill sets and achieved serenity. The Mighty Mainframe with its attendant
chorus of Magi has been relegated to the Back Office. The People have had
computing placed in their hands and they are not going to hand it back to
the Cult of COBOL; the young are downloading courses in programming and
driving robots, writing games, and taking to IT as ducks to water. They
understand Objects, Databases, standard software, exchange of information,
the Internet... All the things that bewilder the old Priesthood.

It really doesn't matter whether Gartner are right or wrong, the world will
vote as it always does...with its feet.