The Greying of the Mainframe Elite 701
bobcote writes "The Boston Globe is running a story about the maintainers of the mainframes getting older and facing retirement. One of the problems is that many computer science programs don't include mainframes in their curricula anymore. From the article: "Amid concerns that America doesn't produce enough technically trained young people, mainframe computer users and developers are especially concerned. Most computer science students concentrate on small-computer technology, such as Microsoft Corp.'s Windows operating systems, or the popular alternatives Unix and Linux. Few have been trained on zOS, the operating system that runs IBM Corp.'s massive mainframes."
No need to register... (Score:5, Informative)
Who'll mind the mainframes? [boston.com]
I wouldn't worry so much... (Score:4, Informative)
You know you are only in school for a few years, but on the job training goes on your whole career, like 40 years or more.
Very little of what I learned in school is applicable to what I am doing now.
Personally I don't think schools should even try to teach such technical skills, leave that for on the job learning or for post college certification training. What colleges need to do is teach people the ability to learn on their own, to have the confidence and the habits needed to go after new fields of knowledge.
That's why I can't stand it when I see universities teaching Java and C#. By the time those kids get out of school that train will have left the station. Maybe teach that to final year students so that when they do their internships they have the basic skills. Otherwise I would expect someone who is really interested in computers to be playing with all that stuff from when they are much younger.
IBM IS training... (Score:5, Informative)
IBM'ers show up at every conference and present. They are easily accesible. I went for the UserBlue AIX specific portions (and got access to network device driver engineers!), but if you go to the non-AIX,non-eServer HACMP stuff its a whole world of applied mainframes.
There is a community out there and IBMers are looking after it.
*nix Admins Are the Best Hope (Score:3, Informative)
1. Strong memory to be able to know which command to use in which context
2. Thorough understanding of logic (this stuff started on mainframes where logic was impreative)
3. Organization. You can't properly admin a *nix box if you don't keep yourself organized. The same applies to mainframes. Windows doesn't really prepare people for this kind of thinking.
Having worked on a VAX and a few Alphas running OpenVMS, I can say that the underlying concepts between mainframe OSes and *nix aren't as far apart as Windows is from mainframe OSes.
Re:IBM should be training? They do! (Score:5, Informative)
You mean something like this [ibm.com]?
IBM Learning Services have a large selection [ibm.com] of courses available for z/OS.
I do think that making these courses better available and better publicized to college students would be a great idea though...
[disclaimer: I work for IBM tough not in the z/OS area. Above is purely my personal opinion]
You're not getting it. (Score:5, Informative)
I was writing the front end to the banking system, first as a VB3 app and then as a web app (in 1996!). As such, you'd run "jobs", basically like how you'd call a stored procedure, and get back the value. So I'd run the job, and before I had taken my finger off the enter key, the result was sitting on the screen.
I asked a "little-old-lady" who was days from retirement how it cached the person's value, and how it took into consideration interest, atms, etc. She told me it didn't. It started from the top of the vsam file, and added and substracted for that person till it got to the end. Then it gave you the answer.
It did this every single time.
I have never ever ever seen anything that could match that machine for raw IO processing. Add to it the fact that it was used by several thousand people all over the world, *and* it ran VM so there were two identical MVS operating systems, then CICS, then the apps....
To be honest, I never got the hang of how to even move around in CICS, but I will give mainframes a lot of credit...when you need to shovel a *lot* of data around, there's probably nothing better.
The fact is people...mainframes are computers answer to gravity...you never see them, barely acknowledge their presence, but you'll miss them when they're gone, because they're the only machines that can handle the staggering loads that would make a cluster of *anything* weep.
Business opportunity (Score:3, Informative)
Re:zOS (Score:3, Informative)
Doesn't anyone recognize the truth when they see it? z/OS is UNIX95 conformant. (I'm not sure about UNIX98.)
Re:No need to register... (Score:5, Informative)
My work is outsourcing most new hires just for this reason, its cheaper to have a vendor do it, and then blame the vendor when things dont get done.
Our HR department cant hire sys-admins at the companies new lower pay scale, so they have been trying to get helpdesk people to move over. Problem they have, new hires make 30-40K lower than everyone else, and expected to do the same job. Soon as they learn enough, they move for higher pay. Turnover and continous training of new people makes it hard on the older more experienced sys-admins who finally end up leaving for a startup or another company without its heads up its ass.
We lost most experienced people, except the real old timers (like me) who been here 7+ years who are just waiting for the lay off notices when the company goes tits up due to piss poor management. I could use 6 months vacation on unemployment. (-;
Re:IBM should be training (Score:5, Informative)
The documentation and source code are (or were) revenue generating portions of the business. If your company doesn't pay for them, they don't get them. In turn, this created some of the most exhaustively complete documentation in the world. It is (was?) a thing of beauty.
Re:zOS (Score:3, Informative)
That depends on your definition of "unix like."
From a system administrator's point of view, you're absolutely correct. I have, on the other hand, written POSIX-compliant C code and seen it compile and run on z/OS with no problem.
Re:IBM should be training (Score:2, Informative)
As one of those retiring...... (Score:5, Informative)
Re:Here to Stay (Score:2, Informative)
That turns out to not be the case. Until a few years ago, the most popular credit union software ran exlusively on MPE/iX on HP3000s. HP has EOL'ed that OS, and you haven't been able to buy an HP3000 since Halloween of last year, I believe. I had to laugh at the date. I've heard systems guys who were obviously on a customer support call talking Unix as they left my bank.
That credit uniion software has been ported to HP-UX, a Unix variant. And in fact, the HP3k (MPE) systems could be changed to HP9k (Unix) systems with the replacement of one chip.
Unix variants can run some huge systems, after all. To stay within the HP realm alone, look at
at Superdomes http://www.hp.com/products1/servers/scalableserve
Personally, I rather think that the days of the proprietary Unices are numbered as well. HP certainly doesn't seem to putting much energy into HP-UX itself these days, and hasn't since the early days of 11i. But they are adding lots of Linux compatability software to the OS, from bash to complete Open Source applications.
My personal prediction is that given the steady capability growth of Linux, and the addition of more and more enterprise software (advanced filesystems, backup software and the like), Linux will eventually subsume both of these markets.
The limiting factors are very probably:
a) Stability. In these realms customers want stability above all else. That applies to the hardware, the operating system, the application, and the roadmaps.
b) Cost. The time and expense of porting applications is probably the most important here. The expense of rare admin talent will be a contributing factor, to be sure. But a comparatively small factor.
This is an ordered list. Cost is definitely the lesser issue, for the vast majority of these business users.
Re:Why should we care? (Score:4, Informative)
Check out Hercules [conmicro.cx].
B.S. (Score:2, Informative)
(http://www.abet.org/Linked%20Documents-UPDATE/Cr
states that the student must be exposed to a variety of systems and languages (and that they must become proficient in at least one programming language). Computer science isn't very interesting or beneficial if the "computer" (i.e., specific computer system running a real OS and applications) is not there. If you attend a 4 year, accredited computer science program in the US, you get MUCH more than what a vocational school would give you.
Well, while part of the problem is the companies (Score:4, Informative)
One thing to help is to get experience while you are in school. Get a job doing something tech related. Maybe it's a basic tech support job that pays $6 an hour to help English majors find the start menu, but it's work experience and it helps. Maybe contribute to some OSS projects as well. You'll find that you can advance even on those campus jobs. Freshman year you are help desk, sneior year you are doing DB develoment for the department's website.
So I think we have some unrealistic expactations from both sides. Many employers think that they should be able to get employees with lots of skills that need no training, and not have to pay for it, but many prospective employees seem to think that a degree should be enough to land them a great job.
Re:But... (Score:4, Informative)
Internships. I make more money than I would ever publically admit to and I blame it all on my college internship. You work for peanuts, or even free, but you gain all that useful on the job experience. Some do it part time and continue to take regular classes, some do it full time for a semester or two. Usually you can earn credits for the work too.
If you are smart and get in the right internship, you can shave 5+ years off your after-college-earning-curve. If you are lucky, you can find the right niche and really exploit it to the hilt.
Re:Make zOS free as in beer. (Score:2, Informative)
Re:Want your own mainframe?! (Score:3, Informative)
No need to be snooty. (Score:2, Informative)
Insightful my eye. More like someone who wants to feel superior to others. The germans have a two-path track to education. You can either take an academic path, or you can take their equivalent of a trade-school path. One side doesn't look down on the other. The same can't be said of some other countries.
It isn't necessarily great for veterans. (Score:5, Informative)
General industry experience isn't valuable enough to obtain even an introductory interview, and one mainframe platform doesn't translate to another in an employers eyes even if the languages and core concepts are fairly similar.
There were a few exceptions, but not very many.
Re:I left the mainframe world... (Score:3, Informative)
So if you were offered a chance to debug Linux kernel code for money, that wouldn't be attractive to you at all either, I guess?
If you're working in the Real World, on mammoth aggregations of code that have evolved over decades, you cannot avoid "debugging other people's code".
Get over it. Despite a CS degree from Stanford, you're just not that special.
If you were, you would strike out on your own and create a new industry or market niche.
Sorry if that sounds harsh, but debugging other people's code is in many ways much more intellectually challenging than producing your own monsters for others to debug.
Quite possibly the reason the people were reticent to teach you anything is that you wanted to be taught, instead of learning. There's a considerable chasm between those things. Another possibility is that they were never informed that they were supposed to take time away from doing the work to nurse the newbies along.
In my experience, IBM documents things reasonably well -- so much so that a major challenge is learning to search the plethora of manuals for the particular clue one is looking for. Start with the Principles of Operation to understand the hardware. IBM Redbooks are sometimes a wealth of how-to info that is generally unavailable. I suspect that if asked, any of the older guys could have given you the view from 40,000 feet, which isn't much, but at least orients you so as to permit intelligent self-directed education from that point forward.
And there are some good texts available -- not many, and they're OLD, but they present a good view that's a lot closer than the view from 40,000 feet. Try Operating Systems: A Pragmatic Approach by Katzan (ISBN: 0442247389) or Systems Programming by Donovan (ISBN: 0070176035) or Invitation to MVS: Logic and Debugging (also by Katzan, ISBN: 0894330810).
Also, there is a wealth of helpful web sites out there, start at Planet MVS [planetmvs.com] or MVShelp.com [mvshelp.com].
And for the truly dedicated, install a mainframe emulator [conmicro.cx] and an old copy of a mainframe OS that's in the public domain onto your PC and debug THAT!