Mainframes still relevant? Career question.

V

Volume

2[H]4U
Joined
Jan 17, 2004
Messages
4,010
The CS department at the state university I attend has a strong relationship with IBM and some other large companies, to whom they send quite a few graduates. A lot of the work they end up doing has to do with mainframe programming and maintenance. The CS department's selling point is that the previous generation of mainframe guys are retiring and there is a large need for people with that skill set.

Are mainframes still relevant today, or are they being phased out? Is accepting a job in mainframes a guarantee that your career will stagnate and that you will become obsolete in the coming years?

My questions are based on assumptions and conversations I've overheard, so I'd like to get a take of someone who is close to that industry.
 
lol

Mainframe sales are still shrinking (as they have been since the microcomputer revolution), but will probably be around for a very long time even if only as legacy systems. If you wanted to go into that field, IBM seems like a safe bet.

For a long time mainframe servers/chips had unique features other commodity server chips didn't have like: ECC throughout core and external memory, hot swap memory, processors and i/o, extra error checking inside core, and ability to run multiple processors in lock step each cycle for extra reliability checking against one another. But slowly those features were added to Xeon and other server chips and now are eating more and more into the last strongholds traditional mainframes have.

I don't have any interest in that field, but IMO that's a career someone should be able to do for at least 20 years if they want. Really, it's very hard to forecast what changes may be coming in 10-20 years and almost anything technology related could require a career change at some point. It's not a bad strategy to always keep your options open.
 
Volume said:
Are mainframes still relevant today, or are they being phased out?
Click to expand...

Mainframes are in fact still relevant, yes. Jobs with mainframes aren't going to be the 'hip', 'cool' ones where you get to work with the latest and greatest technology. However, they will pay well, they will have room for advancement, and it's not like a perfectly competent mainframe developer could never transition into other areas of computing.

There's plenty of places who still use mainframes. I've worked at places in the financial services sector and in the education sector who still use mainframes and (at least while I was there) had nothing on the roadmap for even starting to deprecate these systems.

Mainframes are commonly used anywhere you'd like to maintain some kind of accounts or set of records. Banks and investment companies often use them to keep track of accounts and account balances. Insurance companies often use them to keep track of policies. In the healthcare industry, they're often used to store medical records. Universities use them to keep track of student records and enrollment. I'm sure there's examples of companies in these sectors who don't have mainframes. I'm also sure there's examples of sectors I haven't thought of where mainframes are still used. Either way, it's a big enough field that I would say you shouldn't worry about employment.

Volume said:
...and that you will become obsolete in the coming years?
Click to expand...

Even if mainframes and mainframe programming went away, these would not render a competent mainframe developer 'obsolete'. To be good at mainframe work, you need to be good at problem solving, just like you need to for any other programming related career. All the mainframe developers I know also have a very solid background in databases as well, so I don't think any of them would have any amount of trouble finding a relatively lucrative job that didn't involve mainframes.
 
Mainframes? It's all about tablets and phones now. Even Windows server has a touch interface now. :D
 
Someone in our IT group has a small notebook of cobol code the previous guy left to do some basic maintenance stuff on our mainframe. We're phasing it out, but it is still used for a lot of accounting stuff.
 
Cerulean said:
https://www.youtube.com/watch?v=Hcywf9mwF5U
Click to expand...

In reply to that, when you work on a mainframe you can actually physically feel the lower frequencies it emits and it overall sounds like this: http://www.youtube.com/watch?v=FEblqAqaojg

Mainframes are a very niche market. An acquaintance of mine had to relocate from the U.S. to the Middle East simply because that was the only way for him to advance in his field. Prepare to move where the jobs are, and every time you want a promotion.
 
Since ya'll are on the verge of constructive agreement, let me inject a curious and dissenting opinion:

If you take a job for technology like mainframes, you've got a great chanfce of working with far more experienced people. Most of the people working with mainframes have been doing it a logn time. And, most mainframe users are immensely concerned with performance. Given the experience and the focus around performance, you've got an excellent chance of landing a job writing software that's performance critical, and learning how to write performance-critical software from people who are already very good at it.

Sure, you can find those attributes in other positions. But I figure this job offer has a higher percentage of realizing those attributes than most other jobs you're considering.

I'd agree that mainframe machines are in a waning market, but that market is still there for now. If you spend a couple-five years in that market, you'll be poised to enter more interesting HPC markets, and the sky is the limit there.

Maybe that's not what you want to do -- I don't see what "mainframe maintenance" has to do with CS -- but perhaps it's a lead that gets you where you want to go a bit faster.
 
Thuleman said:
Prepare to move where the jobs are, and every time you want a promotion.
Click to expand...

I disagree with that statement wholly. Maybe your friend struggled to find work domestically, but I know plenty of people who've gone very far with mainframe careers. At some companies, you can even start at a shared desk as a mainframe programmer and make it all the way to being one of the VPs without ever leaving the company. I don't envision anybody labeling mainframes as a 'growing' market, but that doesn't mean there aren't plenty of opportunities in the field.

In my region there is definitely no shortage of mainframe opportunities for those looking for them, and I can't imagine there being a terribly large difference between here and the rest of the U.S.. Certainly there are much, much larger areas of our industry, but even niches have ladders. You can find your way into companies in any area of any field where the only way to move up is to move out, and you can find your way into companies which promote from within where someone in a cube can easily end up in a corner office some day if that's the right role for them.

It depends on the person, the place and the employer far more than it depends on the type of work you'll do.
 
Last edited:
mikeblas said:
I'd agree that mainframe machines are in a waning market, but that market is still there for now. If you spend a couple-five years in that market, you'll be poised to enter more interesting HPC markets, and the sky is the limit there.
Click to expand...

We actually have an HPC unit where I work (one Cray and then a RedHat/ClusterWare cluster with ~1500 cores) though I don't work on it. If the OP wants to get into HPC then he should just go straight up into doing that. There are imho decent opportunities for programmers in HPC that don't require mainframe as a stepping stone.

The route into HPC via CS is learning and having verifiable success with OpenMP (or MPI I suppose) and focus on Linux based distributed computing. HPC systems are not nearly as uncommon as 'proper' mainframes and there's a lot of opportunity for "under the desk HPC-ish" computing where people just buy a bunch of quad-socket multi-core machines to process data outside of an actual HPC data center.
 
Thuleman said:
We actually have an HPC unit where I work (one Cray and then a RedHat/ClusterWare cluster with ~1500 cores) though I don't work on it. If the OP wants to get into HPC then he should just go straight up into doing that. There are imho decent opportunities for programmers in HPC that don't require mainframe as a stepping stone.
Click to expand...

Certainly you don't have to go into mainframe programming if your end goal is HPC. Some students will have opportunities to go directly into this field and learn from experienced faculty who are really strong in this area. There are schools where HPC is critical to the research they do, and there's lots of opportunities for an undergrad to make it onto a research group. On the graduate level, I've seen a lot of internship openings for HPC work as well.

But if the thread starter has a well connected school that can easily place him among experts in the mainframe world, he will have a great opportunity to learn how to write performance-critical code well. And if the mainframe route is more accessible for him than 'entry level' HPC opportunities, there's nothing wrong with starting out working on mainframes.
 
Thuleman said:
There are imho decent opportunities for programmers in HPC that don't require mainframe as a stepping stone.
Click to expand...
Opportunities for entry-level hires?

Thuleman said:
The route into HPC via CS is learning and having verifiable success with OpenMP (or MPI I suppose) and focus on Linux based distributed computing.
Click to expand...
Sounds like not. The mainframe guys, OTOH, are not quite as selective.
 
Mainframes are still huge in banking.

My company is currently hiring young people from within to go through mainframe/COBOL training to take the place of retiring programmers. The current generation of mainframe programmers are nearing retirement age quickly.

I personally was lucky enough to be one of these young people. I landed a job as a non-technical Business Systems Consultant and was approached by management to go through "COBOL boot camp". I have since branched into Java and now Java running on the mainframe, Even though I am currently employed as a Web Developer, I am still in a mainframe technology group supporting the web front end to a COBOL mainframe system as well as java batch processing running in USS on the mainframe.

As stated previously in this thread you will get to work with people who have been programming for decades, I feel very fortunate to work with these "old timers" and I take every opportunity to learn from them.

Technologies you will want to learn if this is something you want to pursue.

TSO/ISP2
JCL (Job Control Language)
COBOL
bonus points for Assembler, in my world assembler exists in only a few places but they are generally pretty critical and only a few people can support it.
DB2
Understand mainframe file systems/storage types such as VSAM
NDM
IBM MQ wouldn't hurt
 
It depends on the industry and company size, but many large companies and business critical industries (finance, banking, manufacturing) utilize a lot of mainframes.

The big difference IMO vs. all the companies using mass market hardware is these companies are heavily invested (both $ and software/capability wise) and usually have a lot of them, thus they are not going anywhere anytime soon. There is still a place for the mainframe.
 
Timely news, a.k.a. IBM reports quarterly results:

system z - down 40% (mainframes)
system p - down 22% (POWER based servers)
(x86 servers, chip manufacturing and storage were all sharply down too)

Ouch.

IBM is already considering shedding its chip manufacturing division, and I wonder if POWER is not long for this world.
 
How do those numbers correlate to overall business spending (on technology) for the same period?
 
That's not the right question. The right one is far more complex: how do those numbers correlate to sales at this point in the current generation's sales curve (i.e. sagging in anticipation of a new POWER 8 architecture release), in an increasingly shrinking market for mainframes... as a portion of Q1'14 server sales.

Sorry, I couldn't answer that. :p IDC and others should have server market analysis in a few more weeks for Q1'14. Q1 wasn't bad for Intel or Oracle, so it's probably still bad news for IBM.
 
I don't think there's one right question, as there are many different things to reason about.
 
Also keep in mind a lot of that is NEW installations, while a lot of the work and need is in existing systems.

Our mainframe had been in operation for more than a decade before we recently updated it. Much longer lifecycle than any of our standard server hardware (which is typically a couple years max).
 
Cerulean said:
Could someone point out to me what each component or part of this image is and does ?

https://upload.wikimedia.org/wikipedia/commons/6/6d/Inside_Z9_2094.jpg

There are a lot of tubes and wires. : o

https://en.wikipedia.org/wiki/IBM_zEnterprise_System#zEnterprise_EC12 -- sounds pretty cool ... 5.5GHz hexacore ... then they say up to 120 cores o_O this makes me curious about mainframes and how they are different at a technical/functional/processing/logic level
Click to expand...

a tube is coolant I believe.

Everything in servers can multithread to a rather insane level. Remember when our PC's transitioned from 1 core to what 8 now? We keep getting more cores and so do they.

Cloud computing and various ways to create super computers using a distributed computing infrastructure can do to computing what torrenting does to files. I am a big believer in the classic mainframe dieing in favor of a more distributed environment.

I'm in southwest Michigan if any programmers have leads for me in the chicago/south bend area. I just finished my Associates and about a year from my bachelors degree.
 
You must log in or register to reply here.
Back
Top