mainframe_kdm

1) It depends on the business goals. In terms of cost, there are a bunch of studies, but basically, distributed is cheaper when you start out, but the costs of a distributed environment rise very quickly. z/OS is more expensive to set up a base environment, but the costs rise much more slowly as the workload grows.

2) Depends on the person. And sometimes, they're the same thing; you can look at something that looks like a very silly design choice, but then you look at the history behind it, and discover that 30 years ago, it made a great deal of sense.

5) Because I always enjoy showing this to people: IEFBR14

Analytics, certainly. With the newest mainframe, the z13, computationally intensive tasks can benefit greatly from SIMD. The data you're mining is usually there already, so that makes things much easier. And in general, z/OS/mainframes were the cloud before it became the buzzword. So anything where the data and processing is done elsewhere, and then the results are the accessed on a different device, can certainly be done on z/OS.

Obviously you're not HR, but I might as well try asking you; is the consideration of opening a new position in the z/OS and IBM z Systems product teams made as people apply (internally or externally?), or are they opened up whenever needed and filled later on?

I'm not really sure, but I believe the answer is the latter. Various organizational units (projects, departments, whatever) have some number of headcount, or people that are supposed to be working on/in them, and then people are found internally/hired as needed.

Are you talking about user exits? Or actually recompiling the JES2 source or zapping the load modules to fit certain needs?

Actually recompiling/zapping the load modules. This was long before my time, but there's a fair amount of z/OS, like JES2, that originated outside of official IBM development, and was shared between customers. I don't know if you've ever heard of the CBT, but they were passed around sort of like that between customers, and they would then take the source code (probably assembler, back then), and modify it to meet their needs. IBM took over JES2 support, but the source code was still available. In 1983 IBM announced that it was going to start shipping some components of z/OS as object code only; this is why they started adding exit points, so that system programmers could still tailor things as they needed to without having to muck about in the source code, which was a support nightmare.

Hah! We have this conversation often internally, actually. I'm in your position; I'm 38, so I joined IBM after parallel sysplex, at the end of the OS/390 days. And I agree; some of the more senior people I work with had a huge advantage in being able to grow up with z/OS. But, then again, they had to. There are a lot of things that system programmers used to need to do, like modify JES2 source code, that they really don't need to any more, because we've been architecting the software to allow you to do the same thing in much simpler ways.

As for training, we have a lot of internal courses; I can't speak to what's offered externally. There's also a lot of written documentation, how clear and concise it is differs, as it does sometimes assume some level of initial knowledge.

We certainly hire people straight into development; they then work with mentors and others on their teams to learn the operating system parts that they need to know. Some people then choose to stay and focus on a single part of the OS, while others bounce around more, and work on different areas. And Level 2, design, development, and test are all very different skill sets, so while people do certainly move between those roles, there isn't really a defined path between them that new people are expected to follow when they come in.

As for finding people, not really. Some new people find z/OS daunting; others really want to dig into it. Personally, I like working with something that has history to it, something that other computer software really doesn't have in the same way, because most other OSes don't make the same commitment to backwards compatibility that z/OS does.

There are also the Master the Mainframe contest and some schools also offer access to z/OS images running under z/VM.