In Defense of IBM z (or why IT is like football)

As I was running yesterday, I was imagining a discussion/debate with senior leadership around the role of the IBM z platform in a modern hosting infrastructure for a global enterprise.?Often, the discussions go like this:

Them:?'we need to move all the old z/OS applications to containers in the cloud. This is the future of digital business.'

Us: 'z/OS is just one of the operating systems on IBM z, which is a platform that can host cloud services, and has extensive containerization capabilities'

Them: 'but IBM z and z/OS are old and we want to get rid of all that legacy stuff and use modern technologies'?

And around we go. It's funny because z/OS is just an operating system that runs on IBM z hardware, which is about as modern as it gets.?Just like Windows is an operating system that runs on x86 hardware.?No one is running around saying we should get off Windows or x86, but from an IT era perspective, is there really any difference??Sure, they are a few years newer, but in a world where technical refreshes happen every 18 months, either both should be considered obsolete or neither.

IT has always been a challenge between preserving the old and embracing the new. If we go back 40 years, the IBM system 360/390 architecture was it.?MVS (now z/OS) and VM were it.?Everyone's business ran on this platform, but everything didn't run well.?

Then we all got excited about "client/server" and the flexibility and perceived low costs of moving computing from mainframes in the data center to small servers under everyone's desk. Everything had to go to Windows on Intel boxes, but everything didn't run well there either.

Then there was UNIX, always lurking in the background and with the development of Linux, it clawed its way onto the podium.?There was a brief time when the strategy was to move everything to Linux, but that was also not the right answer.?

?Now companies have mainframe stuff, Windows/x86 stuff and Linux stuff, each of which at one point in time was the 'wave of the future.'?Like waves they crest and recede, leaving IT landscapes riddled with silos of computing that may or may not be doing the job.??Unfortunately, a lot of applications written for MVS and VM were never upgraded.?The OS and the platforms have evolved, but some applications have remained stagnant.

?What I have seen is that IT is becoming more granular and use case-specific, leading towards the latest buzz around 'hybrid cloud' and multi-platform computing. This is in line with what we in the trenches have always explained to the decision-makers: there is no one-size-fits-all computing solution.?Even the engineers of these products didn't design them to do everything for everyone in every situation. They were each designed to do something really well, but between the marketing departments and IT managers whose ideal is a monoculture of identical hardware running identical software, we've often tried to force all of our enterprise computing needs onto a single solution.?

?How many times have we seen this mistake??A new shiny thing is introduced to great fanfare.?Regardless of your problem, this is the solution. IT managers decide to move their environment to the new shiny thing, and bemoan the "legacy" that gets in the way.?Initial success is guaranteed because the first migrations are tailor-made for the new shiny things (called 'quick wins' in corporate speak).?Excitement builds.?This is what we've been waiting for!?The next set of migrations is harder and slower, but we'll get it done.?We'll sacrifice something on the altar of novelty, so it's worth the pain.?Then the IT managers hit a wall.?A lot of stuff can't move to the new shiny thing - not because the stuff is old, or because people are protecting their jobs or lack the skill - but because it simply does not fit and was never intended to fit.?Then the tension increases, with advocates of the new shiny thing accusing owners of older things - that are running quite happily and stably on their platform of choice - of being reactionary or luddites or, the greatest insult one IT professional can lob at another:?a product of the "mainframe-era."

The cycle repeats itself as the new shiny thing gradually becomes older and tarnished, and it will eventually either disappear eventually or find its niche within the larger environment.?That niche was all it was ever designed to do and all it ever should have been asked to do.

bWith the exception of small, newly launched companies who can choose to standardize from Day 1 on a limited set of options, no enterprise can run solely on the new shiny thing.?Even then, interview the CIOs of those small, new companies in 10 years and the IT sprawl, which they so smugly assumed was a problem they would never have, will bedevil them as their new reality.

Let's say No to the Tyranny of the New and stop trying to brand someone else's technology as obsolete or legacy, and start looking at each platform/OS through an unbiased lens that reveals its strengths.?When strengths are mapped to business need in a thoughtful and responsible way, with the recognition that there will never be a single platform, then we don't have sprawl, we have well-planned hybrid computing.?

Which brings me back to z/OS.?Sure, it was developed in the 1960's but it is not stagnant!?Just like our friends at Microsoft, IBM continues to innovate and improve it year after year. Will you throw out all of your Windows servers because they are running an "old" operating system from the 1980s? No, because they are not.?Windows Server 2019 is no more like MS-DOS than your new Jeep is like the army vehicle from WW II.?Having a historic name and a long history of development and improvement is a good thing, not a handicap.?z/OS 2.5 is about as modern an OS as you can get, with capabilities of doing any cloud function with more speed, security and lower TCO than any other system on the planet.?Yes, it takes skilled IT people to manage it, but what system does not??Do you consider Windows or Linux system administrators to be less skilled then those who can harness the power of IBM z??I don't.?Any company's IT infrastructure needs to be in the hands of skilled, experienced IT professionals and no, they are not cheap. But a freshly-minted graduate with a degree in computer science is going to love managing an IBM z environment just as much as he loves managing an array of Windows boxes.??Maybe even more, since even twenty-somethings appreciate a good night's sleep.

I've also heard the argument that a single platform environment is cheaper and easier to manage.?There is some truth to this until you look beyond the data center.?The technical gymnastics that developers have to exhibit to force their applications onto the one platform is not free.?They have to add complexity in order to overcome the inherent limitations in the single hardware and software stack, rather than exploiting the native capabilities of the purpose-designed platform.?So your sys admin only has one playbook, but your developers are adding time and often layers of software to extract what they need from the single platform.?This time would be better spend adding function and capabilities to the business software instead of conforming to a rigid IT landscape.???Multiple platforms and multiple operating systems means multiple system administrators, but these guys do more than push buttons and hiring a staff with different strengths, capabilities and, yes, biases, will actually lead to a stronger computing foundation.

An IT Manager is like a football coach.?He should assemble the best team, comprising players of different experience and skills, in order to rise to every challenge.?A coach would not field a team of 11 rookie running backs - even if all 11 were college MVPs!?That team would not have the depth and capabilities required to win a game.?Likewise, an IT manager should not stock his data center with hundreds of identical widgets running the same software.?This will work well for some portion of his landscape and horribly for the rest. Not because it's old, or developers are lazy, but because it is not the right technology for the job.

Windows on x86 is excellent at front-end, user-interactive computing, and is a great place for entry-level developers to gain experience.?Linux on Power is fantastic at numerically-intensive computing (true story: a friend from college took a job in a Numerically Intensive Computing group - Helen, if you are reading, Potsdam 1985! - and I had no idea what that was.?I still don't but I know the Power platform is the place to run it).?IBM z and z/OS??I/O monsters. If your business application needs to run transactions or data warehouses, this is the place for you.?Oh, and security??None better.?Your data is always encrypted - and I mean always - and no other compute platform can make that claim, no matter how much encryption software gets stuffed on it.?

So in defense of z/OS, nothing that we are using these days in IT to run our businesses is particularly new. Whether a system is 40 years old, 30 years old or 20 years old, it has history, but this does not make it bad, or legacy or any of the other terms tossed about with disdain by people who want to be seen as leading edge.?Leading edge capabilities can be delivered on any OS, and every platform has its strengths, and its limitations.?If IT managers put their best players on the field, and run workload on the most suitable platform, then they can lay claim to the only new shiny thing that matters:?hybrid computing.

The postings on this site are my own and don't necessarily represent IBM's positions, strategies or opinions.