MVS Systems Programmers... A Dying Breed?

Excerpt from the IBM mainframes newsgroup

H o m e


Date: Wed, 2 Apr 1997 12:36:25 +0000

From: John Duffy

Subject: MVS Systems Programmers... A Dying Breed?

Hopefully this will stimulate some conversation in this news group.

Over the past few months, I keep hearing comments that revolve around a similar theme. Comments like, MVS is an apprenticeship trade. MVS is not taught in the Universities. No more MVS trade schools. I've also heard, from IBM'ers that all the old Systems Programmers are retiring. Hinting that there could be a shortage in Systems Programmers in the future. Also that college graduates in Computer Science have no MVS background. Thus creating a massive training curve for someone who has always worked in a PC or UNIX platform and now has to learn JCL? Has anyone seen this type of trend?

So since most Colleges and Universities don't offer MVS course content could this be a problem for businesses who rely on MVS for there day to day business operation? How does this problem get solved? Do businesses see this problem?

I recently attended a Educational seminar at IBM where it seemed like they were begging Colleges and Universities to get back to the mainframe (CMOS) and enhance curriculumns to teach more MVS content. Has enyone else seen this?

I've recently installed OEMVS and sunk my teeth into UNIX. It's nice and there are alot of commands to remember but when you get right down to it, not a real big deal. I attended a presentation on the R390 platform with lots of UNIX folks in the room and every time the instructor mentioned SMPE or RMF or IPS/ICS, the UNIX folks raised there hands asking what these acyronmns meant. I chuckled then realized that these folks have a long way to go.

So how does the industry handle this problem? How do Systems Programmers face this problem?

I'd love to hear some feed back.

John Duffy Sr. Systems Software Specialist Boston University jduffy@bu.edu
Date: Wed, 2 Apr 1997 16:50:59 EST From: Rick Tsujimoto

I believe it's sort of like the huge tanker that has to start a left turn some 5 miles before the turning point. The only thing that may gum up the works is the cost of software on the new CMOS platforms. The vendors have to realize that their prices must be competitive with comparable UN* platforms in order for MVS to survive.

I recently tumbed through a continuing education catalog and noticed that every mainframe-oriented course is removed and replaced by UN*, C++, OO, etc.

I heard the success rate for projects that move the mainframe apps to other platforms is quite low. Based on what's happening here, I could believe it.

The most popular UN* cry for a solution seems to be "reboot".

  *---------------------------------------------------*
  *  Rick Tsujimoto                                   *
  *                                                   *
  *  rtsujimoto@cusa.canon.com                        *
  *                                                   *
  *  tel: 516-328-4554                                *
  *  fax: 516-328-4369                                *
  *---------------------------------------------------*

Date: Wed, 2 Apr 1997 17:57:08 -0500 From: John Wynton

I will try to be brief.

Treat the following as a bulleted list:

IBM's Architecture is evolving to reduce amount of work sys progs need to do.

Downsizing, reengineering, data center consolidation, mergers and outsourcing, etc, etc.

New recruits, comp sci grads, etc are lured by the 'latest' technologies - ie client/server, object oriented, etc.

Comp Sci courses do NOT include MVS.

Mainframe systems constantly referred to as legacy (= they work) ensures a negative impression on likely entry level candidates.

Lack of training offered. For various reasons, lack of time, because of downsizing, no money because it is all spent on the 'latest' technologies.

There are probably other reasons. But I agree with another responder (Rick Tsujimoto) where he he says Training Companies (NOT US) have let their mainframe curriculum wither. Personally, I don't want that to change because our business has never been better, because we remained focused on CICS, DB2 and other MVS topics, and will continue to do so.

I'd love to start an entry level training program for sys progs!

Interesting topic. I look forward to other input.

Cheers,

-- John Wynton Senior Account Manager Themis Education Services http://www.themised.com
Date: Wed, 2 Apr 1997 15:15:39 -0800 From: Bob Halpern

We have now passed the massive movement to manage by magazine. The mainframe is not dead. In fact, with the new CMOS boxes and OS/390 packaging/pricing, large transaction servers are cheaper on the mainframe, with far more integrity (2 phase commit) than UNIX can offer. Suddenly, SMP jockeys are more in damand, but so are highly specialized peformance and assembler coders. The market is growing again.

I am a dinosaur that survived.


Date: Thu, 3 Apr 1997 02:03:55 GMT From: Tom Longfellow

We were talking abou this at work recently. I think IBM is (intentionally or not) trying to replace sysprogs with sysjanitors who will come in and empty a new ServerPac onto the DASD once every 12-18 months. My thoughts are that no matter how much they try to make install and maint easier, *someone* will have to know and truly understand the underlying products. By the way, I still have at least 20 some years before retirement so there are still a few of us 'younger' people left. I just hope that site attrition doesnt leave me with noplace to work in 15 years.

>
> So since most Colleges and Universities don't offer MVS course content
> could this be a problem for businesses who rely on MVS for there day
> to day business operation?  How does this problem get solved?  Do
> businesses see this problem?

Business, looking beyond the next quarter or fiscal year? When will that happen?

>
> I recently attended a Educational seminar at IBM where it seemed like
> they were begging Colleges and Universities to get back to the
> mainframe (CMOS) and enhance curriculumns to teach more MVS content.

They should do what DEC did in the 60's-70's, Give the darn boxes away. I believe that many of unix bigots today were developed in college. Think of it. A ready made crop of grads every year who know nothing but unix. The net effect is a workforce who can see no solution for computing problems that does *not* involve Unix. Motto: "When all you have is a hammer, everything is a nail"

>
> I've recently installed OEMVS and sunk my teeth into UNIX.  It's nice
> and there are alot of commands to remember but when you get right down
> to it, not a real big deal.  I attended a presentation on the R390
platform with lots of UNIX folks in the room and every time the
instructor mentioned SMPE or RMF or IPS/ICS, the UNIX folks raised
there hands asking what these acyronmns meant.  .  I chuckled then
> realized that these folks (unix have a long way to go.

And the amazing thing is, they think Unix is just as mature as MVS

>
> So how does the industry handle this problem?  How do Systems
> Programmers face this problem?

by doing my job the best I can. And being convinced that there are some things that only MVS can do. Unix has many good purposes too. Both platforms have their purposes. The real difficulties arise when one side (or both sides) puts on blinders.

-- Tom Longfellow TLongfellow@Worldnet.att.net
Date: Wed, 2 Apr 1997 22:06:20 -0800 From: Young

I too, have been wondering about the education aspects of this career path.

IBM mainframes, in todays times, seem to be this dying platform waiting to be replaced, by something "newer", "better", and "more user friendly".

It's true, MVS has never had a pretty face, but MVS design goals have been based upon "reliability", "availability" and "serviceability". I hate to use those old (20 years) IBM marketing terms, but isn't this the type of technology that banks, insurance companies, and retail industries RELY upon?

The design criteria for *IX operating systems, although in many respects is technically sound, did not have these basic business needs in mind.

I have managed an MVS installation for 11 years. I have done development in both the *IX environment, and the mainframe environment. I AM biased towards the mainframe environment, and I have run into some interesting serviceability issues.

On MVS, if your program abends, you get a dump. The dump, although sometimes voluminous, usually has what you need to fix the problem. If a program is looping, you can take a console dump, or cancel the address space with a dump. (if you have a storage overlay, it's all yours) Trace table, RTM2WA's, register save areas, and now stacking PC information all helps in pinpointing the problem, without having to wait for multiple failures, or making code changes or patches to trap a bug in the user environement.

Although all *IX environments are not alike, generally speaking, when a program abnormally terminates, you get a machine readable "core" file. This is helpful, but only if the "core" file is examined with the debugger on the exact same release of the operating system. I have only found simple traceback information available from these core files with any consistancy.

Except for SUN (and I can't remember if it was Solaris, SunOs or both), you would be lucky if you can find a facility that comes close to those MVS offers for tracing events (GTF, VTAM, SLIP, etc.).

The number of aids available to the MVS systems programmer (many that I did NOT mention) and the reliablity of the IBM hardware environment, has always made it an easy for me to provide a stable, reliable environment, that can store and retrieve mass quanities of information.

I sleep well at night knowing my bank is using IBM mainframes. It's too bad the education facet of the computing industry doesn't focus upon the strengths of this operating environment.

John Young
Date: Thu, 3 Apr 1997 08:25:20 +0200 From: POTGIETER NICO

System programmers are indeed a breed that seems destined to wither down to limited numbers. I think there will always be sys progs. Even if only in supplier companies. The developers must have bodies like this to develop new features. The "clients" as such will be left with as someone called "sys janitors" to throw in the odd serv pack. I personally miss the good old CBIPO and struggle to install. That was the only way to learn the product. I think the problem comes from automation and the increase demand on "faster, better in less" time attitude of the business world.


Date: Wed, 2 Apr 1997 21:47:47 -0800 From: DLPolzin

My guess is that most large shops migrate the sharpest application programmers into the Tech Service area (Systems Programming). These would be people that already have a good working knowledge of the shop and how things work and how things get done.

Once there, the education is two-fold. Part A is OJT; mentoring from the existing staff, and part B is good old IBM. They still teach everything about their systems and probably have the best staff to do it. There is also a part C, but this requires some additional commitment from the employer. That is, GUIDE, SHARE and IBM Tech Conferences. These user get-togethers are the single most important activity a systems programmer can participate in. The contacts made with other professionals, both users and vendors, is invaluable. Unfortunately there are a lot of employers who think they can spend their money more wisely elsewhere. IMO they are wrong.

Those are not only my feelings, but that's the way I did it.

Dave Polzin
Date: Thu, 3 Apr 1997 11:27:13 EST From: Bobby Herring

It seems like a lot of shops are cutting back on the money spent in this area by hiring consultants, usually retired se's, and not having to pay the benefits. We bought one of the CMOS boxes and apparently signed agreements that said we had to have ESA 5.1, then OS/390 rel 2 installed by certain dates. This gained us some discounts on the pricing, I guess, but puts a lot of short deadlines on the systems area. We then bring in the consultant to get it in by the contract date. With IBM going to the short life of releases and packaging them with OS/390, it helps them to not have to support the old stuff as long but forces us to keep up.

We don't have the luxury of being a large shop. We have three people in the systems area. We support MVS, CICS, VTAM/NCP with a statewide satellite network(Texas), and all the third party products. We don't get to specialize. By having to keep up with current releases, we don't have time anymore to dig deep into the products. Just lay them down, get the bugs out and turn them over. We don't get to use the finer points of the products unless we hear something or learn about them at a class. We also have less time for those.


Date: Fri, 4 Apr 1997 00:11:42 -0800 From: Tom Brennan

When I started doing installation systems programming in 1983, my company had already been hiring 3 system programmers trainees a year for the previous few years. These were people that in some cases had done application programming, and in other cases (like me) never saw a // JOB card before in my life. We had little effective formal training from the company, and managed to learn the trade mostly by struggling and watching the more experienced people. At least in this company, it was an apprenticeship even back then to learn the basics. To learn more than the basics, you had to strike out on your own and risk making some mistakes. The ones who went to the next level were the ones who learned from their mistakes and the mistakes of others.

When I went to school in '78, part of the curriculum was a 360 assembler class (using a DEC simulation). But even in that time-frame most students seemed afraid of assembler. I think I was one of a small minority that even finished all the projects. It was as if many people didn't want to learn the bits and bytes - they just wanted to tell the computer what to do with high-level languages and installation procedures like A:SETUP (i.e ServerPac).

But then maybe we've unwittingly forced IBM and other vendors into this "we can do without the system programmer" mode. After all, how many of us have called IBM with a problem, only to find out it's that usermod or exit we put on? That kind of stuff must drive them crazy, so of course the only way out is to try to develop some kind of standard, easy-to-maintain system, and try to hide the inner works with OCO.

The last few years have shown our management that the client/server environment hasn't been all they expected, so there's been a resurgence in the mainframe. But how long can this last? Sooner or later those little boxes running Unix or NT or OS/2 are bound to take over. I've done some Visual Basic and Windows C programming, and it's amazing what those things can do, especially if attached to a giant server (maybe a mainframe). Someday the small boxes will have the integrity and security of a mainframe (many ideas copied from us of course!), and be able to do the work that we do today with batch and CICS. They really do copy us though - one thing that always amuses me is when I hear that client/server is now headed toward the 'network computer' in order to minimize the maintenance needed on the client end. Didn't we already have that with a 3270 dumb terminal :) ? But then I wouldn't give up my cut & paste or type-ahead for anything!

Bottom line - I think the mainframe has had a short repreive, but it can't last for long. People teaching must already see this, and although they don't know exactly when the last mainframe will be shut down, it's hard to get people into a class that even slightly appears to have no future.

How can a company handle this? Pay more for the few people left that really know the business (I'm not implying I'm one of them), or head toward an environment that is currently being taught, which can be risky in some cases for critical applications.

Tom Brennan Southern California Edison Co.

P.S. If you're looking for installation system programmer training, I'm a very good part-time teacher and I need money for diapers (not my own, I'm not that old yet :)


Date: Fri, 4 Apr 1997 10:06:18 EST From: Bob Shannon

A few comments on the death of systems programmers:

The mainframe has proved difficult to kill. Among the reasons are:

1. Extremely robust

2. High I/O bandwidth

3. Supports thousands of users

4. Cost effective (at least recently)

5. Excellent support from IBM

6. System Security

Although one can surmise that other platforms will catch up soon, consider that Unix is nearly as old as MVS. Some look towards NT as a long term solution, but NT is proprietary (arguably more so than MVS which can run POSIX compliant applications) and Microsoft hasn't demonstrated it can step up to enterprise support the way IBM does. Also given the importance of mainframe business to IBM, IBM will do everything in it's power to keep the mainframe viable. Finally, mainframes will make sense for a long time to come because it is inherently easier to manage a few things than to manage a bunch of things.

The original thread pertained to MVS systems programmers. Sysprogs were a dying breed that is experiencing a renaissance. Even today new and improved tools threaten to eliminate the MVS sysprog, but remember back to the early days of systems managed storage. One of the perceived benefits was the reduction or elimination of systems programmers. To a large degree that objective had been met. The sysprogs are gone - replaced by storage administrators. (A rose by any other name .....).

The issue of sysprog training/education was raised. As one of the MVS Cluster Leaders at SHARE, I'd recommend SHARE as a good place to start. If you have attended SHARE and believe there are some topics we should cover but don't, send me an email. If you would attend SHARE if we offered other sessions, let me know what they are.

Interestingly we used to have an MVS New User's Project, but it was disbanded due to a lack of new MVS Users. Perhaps it should be resurrected.

Bob Shannon shannon@programart.com


Date: Fri, 4 Apr 1997 14:08:20 EST From: Dave Ulrick

For the record, I should mention that our university, Northern Illinois University (located in DeKalb, Illinois, 60 miles west of Chicago), has offered an MVS-centered computer science curriculum for many years. Our computer science students continue to submit COBOL, FORTRAN, and BAL programs to our MVS mainframe for execution (using the MVS TCP/IP FTP interface to JES2). We have classes which teach the basics of JCL and CICS. Aside from some SMP/E in one graduate-level class, we don't focus on MVS systems programming per se, but we do give our students a good start in becoming MVS applications programmers. Some of us, including myself and several of my co-workers, even end up becoming MVS systems programmers. :-)

If anyone is interested in learning more about our Department of Computer Science, I recommend that you check out .

Dave

Dave Ulrick, Systems Programmer Internet: d-ulrick@niu.edu Northern Illinois University DeKalb, IL, USA


Date: Fri, 4 Apr 1997 17:33:08 -0800 From: Steve Samson >> Interestingly we used to have an MVS New User's Project, but it was disbanded due to a lack of new MVS Users. Perhaps it should be resurrected. <<

Before it was New Users, it was the MVS Conversion Project (I think). That name would be very apt as we reach out for recovering UNIX sites.

Cheers, Steve


Date: Sat, 5 Apr 1997 08:49:49 +0300 From: Yishay Yovel

It is dying all right.

I have been an MVS & CICS system programmer for the past 13 years. There are few things to clarify here before we proceed:

1. We, as system programmers, are interested only with the Mainframe Software Platform. Not the hardware. The question is how long will the MVS/CICS/COBOL/DB2 environment continue to exist. The classic mainframe is already gone.

2. The CMOS mainframes are capable of running Unix (MVS Open Edition that seems to interest nobody). IBM announced that they will soon run NT (probably as a process within MVS). A mainframe which is running NT is not a mainframe (from our professional point of view).

3. The main issue is Application Software Development not system programming. RAD (Rapid Application Development) tools are by far better on the PC then anywhere else.

4. Mainframe as SERVER? I believe we can set up a robust environment which will give Cost/Performance ratio better then any mainframe IBM could set up (The non-mainframe systems are stable (Commercial Unix), Fast (SMP machines based on RISC/Pentium processors), got tons of memory, and are connected via fast I/O channels). IBM could not beat such systems even at the high-end.

5. So, Why do we use them? Mainframe are good ONLY for running the legacy code written for them in the past 30 years (which costs more then all the hardware in the world).

6. Look around you. Did you see whole NEW projects targeted at the Mainframe Environment which were initiated in the past 2-3 years? In the sites I have been with, mainframe applications are being only maintained.

7. The real thing that ties an organization to a platform is the application. A mainframe as a SQL server (MVS/DB2) could be replaced with, relatively, small effort by other SQL servers (UNIX/NT/ORACLE). Applications written in a certain environment, and functioning, will last to its expected life cycle (7-p10 years) no matter what technology will emerge (simply, application environments are NOT compatible in any sense).

8. Cheer up. The Mainframe with MVS will probably bury us all. However, If you are looking for interesting work (rather then financially-rewarding one) you will have to look somewhere else. It is rather quite in the MVS system programming department .

Yishay Yovel.


Date: Thu, 3 Apr 1997 15:00:59 -0800 From: Craig Otway

Bobby Herring wrote:

It seems like a lot of shops are cutting back on the money spent in this area by hiring consultants, usually retired se's, and not having to pay the benefits.

I recently made the overwhelming switch from developing (CICS,DB2 and COBOL) into systems. OJT was something my mentor said would be the most valuable tool he could provide. I said, "buy me an IBM education card", so on my year anniversary that's what I got. With only three systems people including myself I have to become an expert in all the areas that ESA/ OS390 has to offer. I can't wait for 5/10 years of head crashes, DB2 troubleshooting, CICS performance issues to mold me into a "SYSTEMS MAN" or can I? One thing that has become very clear during the last 15 months is that you do not learn systems in the short-run. The classes that IBM offers are very good snap shots of what you need to know; being able to apply what is taught requires an understanding that can only come from hard knocks. As mentioned by another thread to this topic, management/society does not like long term solutions to short-term budgets. It might become the responsibility of each shop to develop this talent or is that the original question? Sometimes I wonder if all this effort will provide for a long career in a downsizing market?

P.S. What is systems programmers? I bet that position has changed over that past 20 years. --

Craig Otway MVS/S390 Technical Support Education Service Center Region 10 400 E Spring Valley Richardson, Texas 75081 972-231-2582 x456


Date: Sat, 5 Apr 1997 21:37:02 CST From: "Edward(Ed) J. Finnell,III"

Death and taxes? So what, guess us moldie-oldies have a slightly different perspective from the a:setup crowd. It's just a stage of maturation of insecurity and self-doubt that's probably better addressed on some psyco-babble list. Just go with what brung you! Inquisitivness, ingenuity, flexibility, determination. Things change, people change-some for the better some for not. Industry changes as trends changed, right now it's the day-to-day bottom line stuff that's in vogue so the market just dropped 1600 points-the pendulum will swing the other way. 499 of the Fortune 500 use MVS(HP doesn't). Would you use your Porsch to haul fire-wood? Depends on how cold it is. Where's it going? Don't have a clue but I do enjoy the ride(no Nissans either)!

EDWARD J. FINNELL,III(EFINNELL@UA1VM.UA.EDU) MVS/Proj. Mgr. http://www.ua.edu


Date: Mon, 7 Apr 1997 11:26:26 +0100 From: "Mike G. Kalinsky"

I don't want to burden the group again by telling you that I am looking for Systems Programmers but the gloominess of Yishay's message struck a chord. I dare anyone of you to put your resume on the Internet and await the phone calls. I am working on requirements that involve new development as well as maintenance. The people working on these project are in some case making close to six figures. I just left a company that claimed the world of Mainframes is dying like that of the dinosaurs. As someone working on 15-20 openings every day for experienced mainframers, I beg to differ.


Date: Mon, 7 Apr 1997 10:03:00 -0700 From: "Sissoyev, Michael - TPMXS"

Thanks Mike,

I myself am experiencing a difficult time finding 'mainframers' to fill the staff. Our 'old, defunct, dinosaur' mainframe is growing and we are preparing to add additional S/390 MIPS (50% increase over the next 9 months). Based upon what we have learned until now, it would be difficult (and very,very, expensive) to replace mainframe capacity, scaleablity and performance with alternative platforms and still mantain a high performance environment supporting hundreds of concurrent users.

Systems Programmers are not dying (unless they themselves decide to) but are transcending the 'mainframe software only platform' to include hardware technology and directions. Operating systems such as UNIX are not replacements to MVS but are complimentary operating system environment which, like MVS, serves to satisfy business requirements. Data centers of the future could easily house both UNIX, NT and MVS systems, all interconnected to share data and the new breed systems programmers will need to understand more than just MVS.


Date: Mon, 7 Apr 1997 13:03:38 -0400 From: Craig John Harris

Register my vote for the resurrection of the MVS New User's Project.

I am not long out of college where I was brought up on the other side of the fence (C, VB, UNIX, MS Windows). I am now in the position of needing a good grasp of MVS but I have found it almost impossible to understand the MVS arena due to lack of introductory material.

I would welcome any initiative that enables the new college graduates like myself to try and fill the shoes of those who are moving on.

Craig Harris craig@waters.net


Date: Mon, 7 Apr 1997 17:41:24 -0500 From: Neil Stenlake

This sounds like a good point in the discussion for an IBMer to 'step up to the breach'. I seem to remember many (too many!!) years ago several manuals, or were they SRA books, that were aimed at people just like Craig Harris. They were introductory in nature, and covered Hardware, MVS basics & JCL, access methods etc...

Perhaps one of the IBM readers of this list may like to research what is currently available - and share it with the list. After all, I am sure it is still in IBMs best interest to foster MVS education.

Regards - Neil.

Neil Stenlake - Technical Consultant Capacity & Storage - Westpac Banking Corp Sydney Australia


Date: Tue, 8 Apr 1997 14:25:51 -0400 From: Tom Hylton

An introductory source...

I received Robert H. Johnson's "MVS: Concepts and Facilities" as soon as I stepped through the SP door. My first assignment was reading it cover to cover. My copy is a little dated now (1989) but still contains a lot of useful introductory concepts. There's probably fresher releases out there, and it's a decent place to begin the learning curve.

If MVS is dying, why are there so many job listings for MVS systems guys in my NASPA Technical Support magazine each month? And before you jump ship, consider that jobs listings indicate (to me at least) that to make 6 figures in one of the other client-server environments, you have to be an executive type and actually run the IS department or at least manage a fairly large number of people. Whereas to make 6 figures in the MVS world, you just have to be very, very, very good with your technical skills. If you're anything like me the prospect of managing others holds no interest, and the nuts and bolts aspects of the job are what draws you to it. So if you're technically driven, MVS seems to offer the most lucrative rewards for that pursuit.

Don'tFlameMeIDontWannaFight,

Tom


Date: Wed, 9 Apr 1997 03:56:51 -0500 From: "Jim Nelson --- jt(ee)n"

I also claim eh???? I have the ICSS server runing under OS/390 Ver 2. and it will beat the pants off of *ANY* unix box Ive seen. Weve got some *HUGE* sun boxes in house, and this little 9672-RC4 probably can out sent any of them. Also, what about the Lotus Notes/Domino port for OS/390??? Even if IBM's claims 1/3 as good as they seem to be (10000 notes users-->to 3000 users), the biggest notes servers we have, start sweating if they get over 500 users cranking away on them.

And as a side note, this OS/390 Ver 2. system is running under VM/ESA!!! (Isnt that neat??? A V=R guest beating up on a stand-alone SUN box...)

Tanx....(just my 2 cents worth on this old and dying thread)

Remember, your fathers mainframe ain't what it used to be!!!!!!!


Date: Thu, 10 Apr 1997 00:02:17 -0400 From: Art Celestini

This one deserves a blow-by-blow:

On Sat, 5 Apr 1997, Yishay Yovel wrote:

>It is dying all right.

Not from what *I* see. I just returned from a conference of MVS software vendors. It was the largest attendance ever, partly due to the added UNIX software vendors who are anxious to port their wares to OS/390.

>I have been an MVS & CICS system programmer for the past  13 years.
>There are few things to clarify here before we proceed:
>
>1. We, as system programmers, are interested only with the Mainframe
>Software Platform. Not the hardware. The question is how long will the
>MVS/CICS/COBOL/DB2 environment continue to exist.
>The classic mainframe is already gone.
>
>2. The CMOS mainframes are capable of running Unix (MVS Open Edition
>that seems to interest nobody). IBM announced that they will soon run NT
>(probably as a process within MVS). A mainframe which is running NT is
>not a mainframe (from  our professional point of view).

OE interesting nobody? See above. OS/390 running NT code? One of the strengths of a mainframe is the ability to run a variety of workloads -- and, all at the same time. I recently heard a story about a computer facility where the mainframe (now CMOS) had shrunk down to a corner of the room, only to be replaced by a farm of disparate "servers," and the customer was pleading for help in sorting it all out. It seems IBM is going after that business by moving the server work onto the mainframe, where the customer knows it would become manageable.

>3.  The main issue is Application Software Development not system
>programming. RAD (Rapid Application Development) tools are by far better
>on the PC then anywhere else.

I write mainframe *assembler* code and I LOVE to do it on my PC. The platform is perfect for the task. But, of course, the code only runs on the mainframe.

>4. Mainframe as SERVER? I believe we can set up a robust environment
>which will give Cost/Performance ratio better then any mainframe IBM
>could set up (The non-mainframe systems are stable (Commercial Unix),
>Fast (SMP machines based on RISC/Pentium processors), got tons of
>memory, and are connected via fast I/O channels). IBM could not beat
>such systems even at the high-end.

I've heard similar statements before, but I've yet to hear about any UNIX hardware or software vendor who has successfully replicated an MVS environment with a full mix of transaction processing, batch, database, time sharing, FTP, web-serving, etc. workloads all running at the same time, and has even come close.

>5. So, Why do we use them? Mainframe are good ONLY for running the
>legacy code written for them in the past 30 years (which costs more then
>all the hardware in the world).
>
>6. Look around you. Did you see whole NEW projects targeted at the
>Mainframe Environment  which were initiated in the past 2-3 years?  In
>the sites I have been with, mainframe applications are being only
>maintained.

IBM's installed base of mainframe MIPs increased by 49% last year alone. I know that the "legacy" [read "they work"] systems have had increases in the amount of data they need to handle, but I understand that is more 10% to 15% per year. So, *somebody* must be writing new code to require all those new MIPs.

>7. The real thing that ties an organization to a platform is the
>application. A mainframe as a SQL server (MVS/DB2)  could be replaced
>with, relatively, small effort by other SQL servers (UNIX/NT/ORACLE).
>Applications written in a certain environment, and functioning, will
>last to its expected life cycle (7-p10 years) no matter what technology
>will emerge (simply, application environments are NOT compatible in any
>sense).

At the high end, I think the thing that ties an organization to a platform is the life of its business. For decades, MVS developers have been guided by the concepts of RAS -- Reliability, Availability and Servicability. The culmination of their efforts is OS/390 Parallel Sysplex -- *designed* to run 24x365. The people running large companies -- and those running smaller companies that they want to grow into large companies -- just don't want to bet their business on anything less. The "I don't know what happened, let's just re-boot" approach to problems that seems to pervade the UNIX/NT community just doesn't measure up.

>8. Cheer up. The Mainframe with MVS will probably bury us all. However,
>If  you are looking for interesting work (rather then
>financially-rewarding one) you will have to look somewhere else. It is
>rather quite in the MVS system programming department .

As a software developer, I've written programs for Windows using VB, C, and C++. After a while, though, these languages get pretty b-o-r-i-n-g. On the other hand, 390 assembler never seems to let me down when it comes to offering some intellectual stimulation of one kind or another.

>
>Yishay Yovel.
>
==============  ,,  ===========
Art Celestini   @@-   adc@galaxy.net (home)
Wyckoff, NJ     <     celestini@platinum.com (work)
PGP Pub Key at   -'#  IBMMAIL(USRWNZEN) --> MVS (work)
pgp.ai.mit.edu  ###   Favorite Site: www.tezcat.com/~smokers
==============      ===========
Warning:  Dates in Calendar are closer than they appear.

Date: Thu, 10 Apr 1997 22:40:00 +0800 From: Bruce Hewson

Hello All,

I thought it was about time I put in my penny's worth (except where I am and have been over the last few years the penny has really died out).

Been sysprogging since 1980 - contracting since 1988....never been out of work...it is still there if you look..

But over the last four years have installed 5 CMOS boxes, and been involved in installing OS/390 in three different sites. In one of the sites we replaced a very aging UNIVAC system!

Currently employed in a site where they claim to have grown from "5 MIPS" to "550 MIPS" in the last 6 years...they started on DOS and became MVS very quickly...still growing with very strong growth forecast....all on MVS..

I have found the support for the CMOS boxes paired with RAMAC style DASD and cartridge tape units to have been the renaissance of MVS shops. It must be the removal of all that weight of copper cable from machine rooms.

So here is my vote that MVS sysprogs will continue to have a very strong career ahead.

Best wishes to all the pessimists.

Bruce Hewson

working in Singapore.


Date: Thu, 10 Apr 1997 23:07:55 +0800 From: Bruce Hewson

An extra point..

I have stopped using the term LEGACY whenever I speak to management or clients....I now use the term HERITAGE....which in my area has a poitive aspect while LEGACY has a negative aspect.

So I say that MVS / OS/390 is the best place to run your HERITAGE systems

Regards

Bruce Hewson


Date: Thu, 10 Apr 1997 08:08:39 -0700 From: "Wayne L. Beavers"

In 1978 Gene Amdahl pointed out in a presentation to employees that people were predicting the death of the mainframe because of the wide spread use of mini computers (remember those?). Gene's observation was that these doomsayers failed to understand that mini computers could collect data faster than people could (terminal users), resulting in the need for ever yet larger mainframes to manage the data.

Move a typical mainframe dasd farm to UNIX and see what happens.

UNIX doesn't have RACF and it doesn't have SMS.

-- --- Wayne L. Beavers mailto:wayneb@beyond-software.com Beyond Software, Inc. http://www.beyond-software.com "The Mainframe/Internet Company"


Date: Thu, 10 Apr 1997 23:32:29 +0300 From: Yishay Yovel

Art, Well said, but reality will prove that the world eventually would drop the CICS/DB2 world on behalf of more advanced (if not, yet, robust) environments.

Yishay Yovel.


Date: Thu, 10 Apr 1997 16:29:20 -0400 From: John Wynton

Yishay Yovel wrote:

>
> Art, Well said, but reality will prove that the world eventually would
> drop the CICS/DB2 world on behalf of more advanced (if not, yet, robust)
> environments.
>
> Yishay Yovel.
>

Original messaged snipped.

What do you mean by 'more advanced' and how do you justify your conclusion - that the world will drop CICS/DB2? From the many clients we provide training to, the opposite is the case!

-- John Wynton Senior Account Manager Themis Education Services http://www.themised.com


Date: Fri, 11 Apr 1997 00:47:34 +0300 From: Yishay Yovel

1. My response to Art was in relation to his response to my initial view about the MVS sysprogs issue.

2. I will try to elaborate on my view.. Basically, I argue that IS is driven by applications and not technology and that the new applications, nowdays, evolve mainly on non-mainframe platforms (in that I mean the software platform not the hardware).

3. The reasons for that are:

a. Functionality (mainframe online appliocations are by far less functional then PC-oriented applications).

b. platform-oriented third party development (their are relatively few software vendors developing solutions for the mainframe (this is changing, to late, with p/390 platform).

3. Take a look at one of the hotest topics of todays IS business: Data Warehouse. Data warehouse tools are targeted at non-mainframe platforms where the mainframe should have been an ideal choice. Why? You can set up powerful environments that are cheaper then a mainframe. Your product will be scalable (you could run it on NTs or various Unix machines and porting should be rather easy). It would not require specialized skills to develop, install and maintain the product in a non-compatible mainframe environment.

4. we could go on like this. My conclusion remains: The mainframe is only good for running its legacy, non compatible, non-portable applications. The new stuff is going someplace else.

Yishay Yovel.


Date: Fri, 11 Apr 1997 00:30:12 -0500 From: "Jim Nelson --- jt(ee)n"

This thread is getting kind of interesting...

Then how come head hunters cannot seem to find enough mainframe people to fill the current employment need.

Also, go to http://www.s390.ibm.com and read the new little blurb from an independent consulting company on the mainframe .vs. unix vs. NT war.

Mr. Yavel wrote something in an earlier thread about I/O speed on these "NEW" client/server boxes. There is nothing (even as we speak) that comes close to ESCON speeds on a mainframe. (Even UNIX consultants will admit that). Go talk to some UNIX consultants once. They do things better then Mainframes, and Mainframes do things better than UNIX....

It can be a happy world...Look at Open Edition...Its happy...(ha ha)

One person's opinion here, of course, but the mainframe has been dead, for about 10 years now. Hmmm...what have we all been drawing a paycheck from for the last 10 years....

Tanx for listening.....jt(ee)n


Date: Fri, 11 Apr 1997 00:38:14 -0500 From: "Jim Nelson --- jt(ee)n"

This thread is getting real interesting...Never tell Mainframe people their dead after they have known it for 10 years. But don't tell the recruters, they haven't found out yet. Or so I have heard. (Check the job ads and see whats hot and whats not)

Anyway, dinosaur's can now do JAVA!!!!! Pretty good for a dead box....

One person's opinion, and its my own....tanx jt(ee)n


Date: Fri, 11 Apr 1997 10:06:09 +0300 From: Yishay Yovel

1. Please put aside this recruting thing.

2. About 10 years ago I was involved with a site running CDC mainframes. That was the only CDC mainframe in Israel and probably there are no more left in the world today. The last person in Israel that was familiar with CDC operating system made a fortune.

3. The inability to hire people is due to the fact that MVS knowledge cant be acquired at the universities. There are very few young people who will go, after graduation, to learn a platform they know nothing about and which the hi-tecxh industry almost disregard.

4. Since some us retire or change careers a gap is created. The dying mainframe is still has to be maintained and since it runs missiom-critical appliocations you would like to have experienced guys running your system dept.

5. The last one to throw the switch on the last MVS box is going to be rich ... and old.

Yishay Yovel.

P.S - I have been a mainframe sysprog (MVS and CICS) for the past 13 years. I am still waiting to be convinced by siginificant arguments and not external factos (like jobs).


Date: Fri, 11 Apr 1997 10:00:04 -0500 From: James Harrison
>>> Yishay Yovel  04/10/97 04:47pm >>>
>2. I will try to elaborate on my view.. Basically, I argue that IS is
>driven by applications and not technology and that the new >applications,
>nowdays, evolve mainly on non-mainframe platforms (in that I mean the
>software platform not the hardware).

Front end applications, maybe... everyone is going GUI.. but back end??? I think not. You use whatever is appropriate for the job. The MF is just another server on the network now... We *just* got DB2!!!! There is no other platform right now that can handle the volume of data we have to deal with... 10 years down the road, the answer might be different, but right now there is only one answer.

>3. The reasons for that are:
>
 >a. Functionality (mainframe online appliocations are by far less
>functional then PC-oriented applications).

See above... front ends are going GUI/PC, BIG data *isnt*.

>b. platform-oriented third party development (their are relatively few
>software vendors developing solutions for the mainframe (this is
>changing, to late, with p/390 platform).

Tell that to CA! (Actually, they bought em all!)

>3. Take a look at one of the hotest topics of todays IS business: Data
>Warehouse. Data warehouse tools are targeted at non-mainframe >platforms
>where the mainframe should have been an ideal choice.

Why? You can Tell that to Platinum & Oracle!

>set
>up powerful environments that are cheaper then a mainframe.  Your
>product will be scalable (you could run it on NTs or various Unix
>machines and porting should be rather easy). It would not require
>specialized skills to develop, install and maintain the product in a
>non-compatible mainframe environment.

In case you haven't looked lately, the MF is becoming very much like those boxes you are talking about above. And, uh, no specialized skills? Puleeeese!!!!

>4. we could go on like this. My conclusion remains: The mainframe is
>only good for running its legacy, non compatible, non-portable
>applications. The new stuff is going someplace else.

Yes, the new user interfaces are going somewhere else... the huge gobs of data, security and stability are not moving anywhere for now....

Jim

*standard disclaimer* just my own opinion.. etc...

back to lurk mode...


Date: Fri, 11 Apr 1997 16:48:12 -0400 From: "Stephen E. Bacher"

Bruce Hewson says:

>
>I have stopped using the term LEGACY whenever I speak to management or
>clients....I now use the term HERITAGE....which in my area has a poitive
>aspect while LEGACY has a negative aspect.
>
>
>So I say that MVS / OS/390 is the best place to run your HERITAGE
>systems

It's only a matter of time before "Heritage" will have the same negative connotations as "Legacy". After all, what could be nicer than leaving a "legacy"? It was considered polite once. - seb


Date: Fri, 11 Apr 1997 13:58:45 -0700 From: Lionel B Dyck

The latest edition of Datamation on the cover asks: Are the Mainframes Cool Again?


Date: Fri, 11 Apr 1997 17:06:33 -0400 From: Art Celestini

A few additional thoughts here, as a kind of a general response to some of the other follow-on posts:

1. I believe the mainframe business will stay vibrant *indefinitely,* simply because of the caliber of the people involved and their commitment to delivering top-quality, competitive solutions. They are the best in the business and *that's* why they make the big bucks.

2. It may seem like the mainframe lags behind other platforms in one technology or another, but a closer look would reveal that once a technology is proven truly worthwhile, it shows up on the mainframe in its most robust form (e.g. fiber optics, RAID, etc.)

3. Yishay responded to my original rebuttal by saying, "...but reality will prove that the world eventually would drop the CICS/DB2 world on behalf of more advanced (if not, yet, robust) environments." There's a dozen ways I could come back on this, perhaps following up along the lines of (2) above, but let's con- sider just one "more advanced" feature that is unique to the mainframe: Parallel Sysplex. It is remarkable technology and since it's now maturing and being deployed in more data centers, the kind of scaleability it provides will become more and more ob- vious. The UNIX/NT world has nothing like it, and I suspect it won't for a very long time.

4. I'll agree with Yishay on the fact that there are a lot more new applications being developed for PCs. I think it's a natural evolution. I seem to remember from many years ago when I did some 3270 work, specifying a "numeric-only" field, so that the terminal did some "data validation" for my host program. Fast-forwarding about 20 years, the PC is the ideal means of offloading the obsequious task of interfacing with the dumb user and validating every little thing he/she enters. If that process involves GUIs and mouse clicks and drop-down lists that make the user a happier camper, then great! What's happening, though (as suggested by Wayne Beavers with his story about Gene Amdahl) is that it's still creating MORE work for the mainframe. Once the merchandise order, medical history, insurance claim, etc. is "clean" at the PC, good business practice behooves moving the data to a more reliable, secure platform. Therefore, most serious new applications developed for PCs include a "back office" component. And, for those "bet-your-business" applications, companies will want the *most* reliable, *most* secure host there is, which, of course, is a mainframe.

5. The biggest problem facing the mainframe industry today is getting it to grow faster. Most mainframes are owned/leased by larger companies, and each year, there just aren't all that many com- panies crossing over into the "larger" category, and buying their first mainframe. IBM, of course, has been taking steps to expand the lower end of the mainframe market. P/390 and Multiprise have brought the low-end cost down to where it's competitive with the higher-end UNIX boxes, but there's still the problem of systems programming, which is two-fold: (1) There aren't enough ex- perienced SPs to go around, and (2) it's hard to justify a full- time SP is a small shop. So, I think the effort underway to make it easier to install and maintain OS/390 is mostly targeted for this market segment. For the larger shops, I suspect the buttons and dials that SPs are good at tweaking will stay there, so they can continue to get the most from their big systems. But, to penetrate the low end, we really need a "manifestation" of the operating system that installs in a snap and runs well right out of the box. Also, while there hasn't been a lot of talk about it just yet, I believe the ISVs have to do the same kind of thing with their OS/390 products, for the same reasons.

==============  ,,  ===========
Art Celestini   @@-   adc@galaxy.net (home)
Wyckoff, NJ     <     celestini@platinum.com (work)
PGP Pub Key at   -'#  IBMMAIL(USRWNZEN) --> MVS (work)
pgp.ai.mit.edu  ###   Favorite Site: www.tezcat.com/~smokers
==============      ===========
"Very funny, Scotty. Now beam down my clothes."

Date: Fri, 11 Apr 1997 16:07:25 -0500 From: "Buschmann, Al"

Legacy implies something that has died and is willed to the next generation. The mainframe is dead, NOT! I prefer the term "CLASSIC" as the mainframe environment has always been of the highest quality.


Date: Fri, 11 Apr 1997 15:05:50 -0700 From: "Wayne L. Beavers"

From the CrossPlex Concepts and Facilities Guide:

"Keep in mind the term LEGACY means something of great value left for others to use in new ways." ---

Wayne L. Beavers mailto:wayneb@beyond-software.com Beyond Software, Inc. http://www.beyond-software.com "The Mainframe/Internet Company"


Date: Sat, 12 Apr 1997 14:32:23 -0700 From: Josef Werner Bvck

There another reason, I think, why mainframe isn't dead and will stay alive long.

It's nice to have GUI front-ends. but with pc and distributed data and distributed programs you have more troubbles in application development than benefits.

you have to care about the location of data and if you want i.e. reports from the whole enterprise, you must gather the data from 50 or more locations (depending on your program organizations).

on the other hand you have as many copies of your program, as pc's exist with all problems we know (distribution of programs, different hardware and software on evry pc, already detected errors)

This situation can be cleared with intranet-solutions. you have one copy of your program, the data you transmit via line is not much more than with classic 3270-applications. you can widen your network of your company without large investments and application development is rapid.

to keep your data and applications in a central place the mainframe is a fine thing. You can drive reports worldwide with not more than 1 SELECT and you have all your data consistent.

in MVS you have all components you need for such solutions (security included), your environment (hardware, software and MANPOWER(a good team is a lot worth )) and IBM makes a lot of efforts to support this market too.

So - i think - the mainframe will go back to a similar situation as with 3270, before pc. Data will be centralized, applications are reduced to nice screens, the overhead of pc-administration will be reduced and the intranet (or internet) will solve some problems, workstations brought us.

and the cost of DP will be reduced. (Cost of dp developed from average 1,5% of revenue in the early 80's to (in some companies) 8 to 10% with workstations and a lot of servers and 1.000's of licences). and cost is the central argument. --

mit freundlichen gruessen / with best regards

Josef Werner Boeck j.w.boeck@t-online.de


Date: Sun, 13 Apr 1997 20:21:01 -0700 From: Steve Samson

All right Art!!

I think the discussion takes a few odd branches. The "mainframe"'s (ugh! how I hate that word!) future is solid for all the reasons you and others have cited. (I prefer to think of "OS/390 platform" as the name for the hardware configuration.) The system programmer, however, is and has been an overhead factor in many installations, nevertheless often doing heroic work to supplement what the stingy operating system vendor refuses to implement [correctly].

In the area I care about, performance management, the need for the sysprog declines dramatically with the implementation of Workload Manager goal mode. Instead of worrying about such arcana as paging rates and MPLs, the person responsible for setting up performance parameters actually has little more to do than talk to the real customers or consult the service level agreement if there is one. As the subsystems sign up for advanced levels of WLM management, the even more specialized CICS and IMS wizards will also have to look for honest work.

The basic work of establishing an efficient system configuration and ensuring that it is maintained as workload grows will continue, but that is not in my mind system programming. It's more like performance/capacity with accountability.

Today's environment just does not tolerate obscure mods or neat hacks that tend to retard currency. Too much is happening in OS390!

Steve Samson (speaking for himself)


Date: Wed, 23 Apr 1997 14:55:38 -0700 From: Skip Robinson >

Schmegacy, schmeritage. Raise your hand if you know the answer to this one. How many Lear jets does it take to obsolete a 747?


Date: Thu, 24 Apr 1997 10:10:06 -0400 From: Duane Jeirles
>>
>Schmegacy, schmeritage. Raise your hand if you know the answer to this one.
>How many Lear jets does it take to obsolete a 747?
>

I would guess around 20. Its been a long time since I've been on a 747 (about 20 years). I think a 747 can transport 400 people and a Lear jet can transport about 20 people. Also, I would guess that a 747 costs around 50 million and a Lear jet 5 million. As you can see a Lear jet is much cheaper. Unless you do the math!


Date: Thu, 24 Apr 1997 11:10:10 -0700 From: Skip Robinson
>>
>I would guess around 20.  Its been a long time since I've been on a 747
>(about 20 years).  I think a 747 can transport 400 people and a Lear jet can
>transport about 20 people.  Also, I would guess that a 747 costs around 50
>million and a Lear jet 5 million.  As you can see a Lear jet is much
>cheaper.  Unless you do the math!
>
>

Not sure how to take your last sentence. This is *not* a math problem. If it were, the world's airlines--who use lots of math to enhance profits--would have converted their fleets decades ago. The Lear jet is a fine vehicle for the purpose its designers intended, which was *not* mass transit. What do you get when you replace a 747 with a bucketful of Lears? An army of client-servers.


Date: Fri, 25 Apr 1997 08:35:54 +0800 From: Goh Boon Nam

I think the anology is not so much comparing a 747 with Lears but rather a Hercules transport plane with a Lear. The IBM mainframe is an industrial-strength utility machine -- good for hi-volume batch and transaction processing but missing many of the frills of smaller machines. IBM has a unique opportunity of turning its Hercules into a 747 but is progressing so slowly and going in all the wrong directions that you may find one day soon that it is the Lear jets that turn into 747's rather than the Hercules.

(What do I mean by IBM progressing slowly or wrong direction? There is was a big gap in the last few years when the smaller machines and their OS's could not compete with the IBM mainframe but IBM did not grab the opportunity to quickly turn their mainframe into the user-friendly, developer-friendly super-servers that they had the potential to be. The status of the mainframe now is that it is only a partial super-server -- you need a slew of additional software/hardware from IBM with an army of systems programmers and IBM consultants to get it working properly and afterwhich the poor developer better have a Phd to be able to understand how to do the low level programming that IBM insists must be done before applications can be churned out. You can't do RAD developement with the IBM mainframe -- at least, not with the mainstream IBM products. Take a simple example - DB2 v5 - it's not yet out and when it's out, it will still only provide very primitive ways of writing stored procedures (I believe the poor programmer still has to manually update the systems catalog to feed it with the meta-data on the parameters that the stored procedure will use); why can't IBM just steal a few ideas from the many competitors that are already in the market - if it's a simple select statement stored procedure, then the DBMS can generate the update itself. MQSeris - why try to make people convert to an entirely new architecture when CICS is already doing up to say 80% of what MQSeris is supposed to do for transactions. (The only justification I can think of in using MQSeris is in EDI-type store-and-forward transacations and even then, only if the comms infrastructure is bad or is a road-warrior remote access type situation which is getting rarer nowadays now that comms infra is becoming so reliable worldwide CICS Client - Here was a good product lying around for years; hardly publicised; the little bit of additional development that needed to be done to make it developer-friendly not done until now and even then, there's still lots of futher improvements that can be made.)

Maybe MVS System Programmers are a dying breed.


Date: Fri, 25 Apr 1997 17:38:53 -0700 From: "Wayne L. Beavers"

Don't confuse being a Web Server for data with a Web Server for applications!

If all you want to do is display data (HTML, GIF, MPEG, JPEG, TEXT, etc.) then it takes 11 minutes to install the software to do that, assuming that you know how to code JCL. You don't need OS/390, you don't need Open Edition and you don't need LE/370 either. All you need is an MVS/ESA Web server. Enterprise Web/MVS can serve most any standard mime type.

Now, if what you want to do is display results of applications on the Web, that is a little more complicated. Running applications (CICS, IMS/DC, VTAM, TSO, ISPF, Roscoe, Wylbur, etc.) that access data in large data bases (IMS, VSAM, IDMS, FOCUS, etc.) is an entirely different problem. In most of these cases the data is going to have to be processed by a CGI program. Some one is going to have to write those CGI programs. You can write them to run on any platform. I recommend that you run your CGI programs as close to the data as possible. For mainframe data bases that means run the CGI on the mainframe and send the html to the browser.

Warning: My comments are very biased. I am one of the Enterprise Web/MVS developers.

BTW my 11 minutes to install was for someone that has installed it before. One of my customers installed it for the first time yesterday and it took him 20 minutes. Several CGI's were in test mode the same day.

--- Wayne L. Beavers mailto:wayneb@beyond-software.com Beyond Software, Inc. http://www.beyond-software.com "The Mainframe/Internet Company"



Back to home page