From Dumps to DevOps: Mainframe-transformation Lessons from Compuware Abend-AID
Since the dawn of distributed computing, there has been what Compuware Software Developer Bob Markle calls “mainframe elitism.”
“We do it this way, we refuse to change because it’s so different,” he said. “Well, it’s not different. We can make the mainframe more palatable, less arcane. We can listen to some of the younger folks, ask them what their ideas are to help us change the mainframe environment, rather than say, ‘We do it one way.’”
“Change” is what Markle was doing at 25 years old when he and his colleague Walt Thomas wrote Abend-AID, the first piece of software able to “alleviate the time-consuming, tedious and manual process of combing through stacks of ‘dumps’ to determine the source of program failures,” according to Markle’s bio under Enterprise Systems Media’s Mainframe Hall of Fame (yeah, Markle’s name means something in this business).
In 1974, he and Thomas were application programmers at Standard Oil, but they were itching to start something of their own. In the summer of ’74 at a cafeteria in Cleveland’s Guildhall building, they had an idea. Under the name Tomark, the duo wrote and sold Abend-AID. By 1977, the product was gaining traction with, shall we say, colorful ads and writeups in notable publications like Computer World.
In that era, companies incurred huge costs when programs abended; costs in paper and in time—CPU time, print time, transmission time, people time. As another Tomark ad in ComputerWorld from 1979 asked, “Why does a large computer system have to produce a 200-300 page dump every time a program has an abnormal ending? It doesn’t.”
“Instead of printing everything, Abend-AID could see how many abends you have, what the size of your memory was, how much paper to guesstimate, what percentage of memory got dumped,” Markle said.
From Dumps to DevOps
Tomark eventually sold Abend-AID to Compuware. Markle went with and continued to innovate the product, until retiring in 2015 after nearly thirty years with Compuware.
During this time, Markle began a project writing C in open-source compilers to navigate the data of some old binary SVC dumps, for no cost other than his time. But he was spending a lot of it on this newfound hobby.
By 2017, he found himself looking for another job in mainframe IT and found his way back to Compuware, where he’s once again helping create modern software that lets the mainframe integrate with and operate like any other platform in an enterprise.
Markle says the mainframe has always been siloed from the rest of IT, but from a technician’s perspective, there are more similarities than differences between it and other systems. The differences are far from major and they are surmountable with DevOps.
“We don’t need to make the mainframe special, just make it common sense, and DevOps is equal to common sense. We can’t work in a vacuum, we can’t pretend to be special. We’re a living organism, so the heart and lungs better work together,” he said.
As a mainframe product that is over 40 years old, Abend-AID is a prime example of how to apply this philosophy to something that may seem beyond the bounds of modernity.
While at its core Abend-AID is still a mainframe application failure resolution and fault management solution, today it’s made graphical and accessible through Topaz Workbench, Compuware’s Eclipse-based IDE. It also integrates with mainframe tools like Syncsort Ironstream and Compuware Xpediter and File-AID as well as with non-mainframe tools like Splunk, altogether aiding the growth DevOps in enterprise IT.
In 1974, Markle helped develop something that didn’t exist for the mainframe. It brought automation, greater visibility and improved communication in one area of mainframe operations. Based on what that product has become over 40 years later, it makes you wonder why anyone would believe the mainframe can’t change.