The classic hits keep coming from IBM: z/OS set for big update in September IBM has teased a significant update to its z/OS mainframe operating system. A preview of z/OS 2.5 appeared on Monday and represents business-as-usual for Big Blue, which announced previous point releases for the OS in late February 2019 and 2017. As with almost everything IBM does these days, the focus is hybrid cloud, …
Why not? It has done its intended job for decades and provided its users with clear upgrade paths to enable them to use modern technology without requiring them to abandon their old, stable codebases, and will continue to do so for the foreseeable future. The downsides of migrating off mainframes onto PC-based server networks are various and well documented, and anyone who attempts to do so without taking the cost of those downsides into account will only invite disaster.
You could do that with LPAR and/or VM/CMS back in the 1980s. Production MVS/XA in one partition; VM/CMS in the other partition hosting any amount of test MVS and test VM/CMS etc. Virtualization three levels deep if you needed it.
And you could do this with a single CPU and something like 64Mb of RAM while maintaining interactive response and batch throughput.
Well, I find it interesting anyway, my experience with MVS (which was renamed z/OS ages ago) dates back to summer 1983 as an operator at Philips in Croydon. Many other readers will go back further than this.
I knew that JES3 was going to get the chop, but it's still interesting to see that it's taken until now. I never encountered it, I always used JES2 and only came across customers (when I worked for IBM 1984-2008) who had JES2. I've run JES2 myself for fun on a PC and on a POWER4 p690 even (on an emulator on top of Linux in both cases).
Just interesting because of the long life of the software which is supported by the operating system.
Just interesting because of the long life of the software which is supported by the operating system.
This kind of lifecycle should be the norm.
For mobile and consumer stuff, sure short lifecycles are fine. But for software that is basically infrastructure then it should be expected to last decades. (Not without maintenance obviously).
There are many buildings decades old that do servicable jobs. If you have complex software that underpins your (now large due to sucess) business then replacing it every few years would be insane.
Now the initial flush of rapid development in IT is over (10 year old hardware is usable today in a way it wasn't in 2000, for example), certain classes of software need to be written to last.
Though I'm a die hard mainframer through and through with a complete understanding of the reasons companies keep on buying this kit, I also know just how much of IBMs revenue, and especially IBMs profit margin, comes from this 7 billion dollar a year business. Without competition, the mainframe has become an exploiter of its unique niche, and in a nasty not a nice way. It's a pity that this is the way the business world for large scale enterprise computing has become, because it is great technology and it need not have become the exploitative monopoly it has.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
