As someone who has worked with what we now know as z/OS since 1980, it’s fair to say that I have an intimate relationship with the architecture. Having started coding CCWs to control a 1287 OCR in 1980 through to delving into the Principles of Operations manual that we know today, I’ve just about done it all cover to cover. I wanted to make it clear that this is an opinion piece coming from someone with a long-term loving relationship with what we now know as z/OS.
What makes z/OS so capable is the z architecture. There’s no doubt in my mind that the capability of z is second to none. I’ll give a real-world example in another writeup that makes x86 seem like the small family runaround vs the z road train. Each fills a different purpose with a different pricing model. But I wanted to dedicate this piece to how it’s time for us to consider life alongside z/OS and to embrace OCP on z.
Running OpenShift on z supercharges the new container based eco-system. Having all of the new container-based tech (previously the preserve of the x86 based world), but able to handle the workloads currently the preserve of z/OS, if it’s a road train that you need. Think of a containerised workload of 40k trans a second running on hardware that costs about a couple of million quid with six 9’s availability. Oh, and data gravity that would compete with any black hole in the universe. As I said previously, more on real world capabilities of z in another one of my rambles. Price the same rig to run “in the cloud” and see how far a couple of million over 3 or 4 years gets you. Good luck with that one!
So, what am I getting at? Well as long as IBM give the impression that a career using IBM is a career using z/OS, then they are still asking people to lash themselves to a mast that, in a lot of large shops, is increasingly flagged “do not invest”. The natural dogmatism that exists in most big organisations, provides the x86 architects with the ammunition needed to outsell z within a large organisation. The perceived “clunkiness” long associated with mainframes just clouds the issue.
Furthermore, in large organisations hype tends to mean “that’s what we should be doing”, but, z/OS cannot compete in the world of hype. The guys on z have never had to sell the value of z because it hasn’t really had any competition for running road train sized workloads. I would say the same is true of IBM until Red Hat came along. But now, some bloke who started out selling books online, I forget the name, has easily out hyped IBM. Add to that the company that nicked the DOS market from IBM (as well as the “windows” PC operating system market) plus some student developed social media app and there’s enough hype dollars to convert me to being vegan, if they were to ever put their minds to it.
So hyping z/OS is detracting from the real star IBM has for running a road train, that is, OpenShift. The fact that it helps our carbon footprint by saving hundreds (maybe thousands) of tonnes of CO2 (emission levies will soon will be used to raise taxes to help us look as though we are serious about holding to the 1.5 degrees of global warming) will look good in the annual report for any large financial institution. With someone smarter than me at marketing, surely z can prove itself as the only way to run your road train sized container workloads.
If I can pivot slightly. I’m not sure making solely z/OS attractive to new starters as a key marketing initiative is necessarily the way for IBM to go. I know mainframe customers have been asking for it, but, in my view, IBM should hype up transferable skills so that people are not going out on a limb to go z vs x86. You can use your transferable skills to work a portal running on x86 but also use them to build and run some of the country’s key infrastructure running on z. Make z the direction to go in with z/OS and OCP both playing a role.
Now let’s not kid ourselves. The x86 based market for development tools is, and will probably remain, the platform for development and functional testing. This means for a new starter that skills in this area are fungible regardless of the deployment platform. OCP on z has versions of the tooling that will be familiar to x86 guys but capacity needed to run these huge workloads.
We should also note that the days of “mainframe changes” and “off-host changes” are fast disappearing. For a long time to come, even the smallest business change will cover both platforms requiring joint coordination from ideation to benefit realisation. Transferable skills will be key during this transition period. As long as major vendors have silos of x86 vs z then the customer base will continue to be siloed too. There’s a chance here for someone to lead the charge into solutions that coordinate business change on both platforms. IBM own z so they have the chance to provide the road train operators a solution.
I hope that this has provoked you into sharing your opinions. I’ll be working on more of these blogs taking a deeper look at how to integrate z/OS and non-z/OS changes into a single delivery pipeline.
Thanks for reading.