SHARE -- Sessions
Wandering around SHARE at the Colorado Convention Center, I have peeked into many rooms, but haven't sat myself down very much. There was a time when I would have found sessions on 64-bit mods to Z Architecture STORE HALFWORD LONG, or on the REXX compiler, or on installing Linux on z/VM entirely swoomy.
That was Then, This is Now. Nowadays, I generally save my finest tech chops for quantum computing and focus my classical computing interest on the human issues, the aesthetics, the epistemology of programming. But there is also fun in programming, and everyone knows, or should know, there is fun in mainframe programming.
Mainframe programming being a mission-critical environment for much of the industrialized world, and faced as it is with an aging workforce and limited interest among young programmers, there are plenty of human issues there.
A placard announced a session called "Planning Your Career Before Taking That First Mainframe Job". Kenneth Tomiak, who has careered around the mainframe world for some decades, offered his insights into the kind of jobs that are out there, the kind of skills and interests they require, along with the challenges and rewards.
A "lunch and learn" session provided me with a veggie box lunch, so I sat and dutifully listened to an IBM exposition of their concept of cloud computing. You have to hand it to IBM corporate culture, they are among the most capable presenters of the top-down, business-oriented view of a wide-swath trend or innovation, and presenter Andrea Greggo was no exception to the rule.
If you have encountered cloud computing, you won't be surprised to learn that it consolidates hardware, focusses customers on services rather than infrastructure, and that its largest operational cost for the service provider is the personnel involved. If you have ever heard of IBM, you will not be surprised to find that IBM feels mainframes have a place in the infrastructure of cloud computing.
Our old friend Rob van der Heij of Linux on the IBM S/390 fame presented a session about Linux on z/VM memory management. "There are some challenges; not everything is automatic," we were warned. Linux is rather piggish by nature, it is used to being the resource manager and does not go supine in the presence of the z/VM hypervisor. On its part, the z/VM hypervisor can't peer through Linux's native virtual memory and task management to detect which memory pages are more important to retain in main memory and which pages to harvest.
Rob possesses an exemplarily detailed technical intellect, and he's as expert as you could wish, having been part of the original port of Linux to the mainframe. He's also a wit: when I mentioned in conversation the failure of LIGO to detect gravity waves and its potential bust of General Relativity, Rob opined that the universities would be compelled to issue a recall of physicists.