IBM mainframes still matter for much the same reason Willie Sutton robbed banks — because “that’s where the money is.” That may be an anachronism, however. If Willie Sutton was alive today, would he still be robbing banks?

In a previous blog I took a look at the early phase innovation of the iPhone using physics as our guide. In this blog I’d like to look at the IBM mainframe (software specifically) through a calculus lens. And what better guide for late phase innovation than IBM — the typewriter company from 1933 to 1991. How have they fared staying relevant in today’s fast paced, information-obsessed world?

In particular I thought we’d borrow three calculus concepts:  limits, differentiation and integration.


Where are the new mainframe start-ups?

In the timeline of data processing, the IBM mainframe (IMS, CICS, DB2 and even z/OS) represents an infinite series. And the incremental improvements over the years have served IBM well. The reliability, scalability and security of these subsystems are ideal for today’s high-volume transactions driven by the Internet and mobile devices.

This leads to an obvious question. If it is truly so well positioned for today’s world, where is the customer validation? Where are the new mainframe organizations? Where is the startup that decides to go the mainframe route?


differentiation z/OSHas z/OS kept pace with the demands of today’s fast changing world? This is hard to gauge since IBM marketing moves at a much faster clip than IBM technology.

IBM provides a wide variety of access points from open systems to the mainframe — stored procedures, MQ, CICS Transaction Gateway, etc. If an application wants to interface with mainframe systems, the plumbing is in place.

But, while the story is still evolving, IBM mainframe has a tentative relationship with newer technologies:

  • z/OS does Java poorly (so poorly that IBM doesn’t even charge you for the processing!);
  • zLinux is a very attractive and very green alternative, but can’t seem to reach any level of critical mass; and
  • Hadoop? Well let’s just say there are philosophical differences to be overcome.


integration z/OSFinally, has IBM mainframe successfully created a welcoming big tent for programming talent?

Again the grade is incomplete. For years, decades even, IBM mainframe has been incredibly insular — unique programming languages, unique development environments, and a unique programmer society. These factors have led to a unique and troubling phenomenon — the aging of talent to the point of putting businesses at peril.

To IBM’s credit, they are working frantically to address this issue. And for once the wind may be in their favor. The current generation of programmers is not nearly as wedded to languages and platforms as their fore bearers.

Does this all matter? I say yes. IBM still exerts a significant gravitational pull in the universe of enterprise information systems. However, IBM mainframe is big enough to plow its own path in some areas, but no longer so big that it can ignore the rest of the industry. Which begs the question, is the IBM mainframe at risk? Of course. Everything is. Just ask Lehman Brothers or Pontiac. In some ways the hardware has similar issues as the programming community: so insular that it relies on a finite pool of users, which keeps shrinking via mergers and acquisitions.

In my opinion, IBM should lead when possible, follow where appropriate and, consider moving out of their comfort zone into areas like open source, to better attract today’s software entrepreneurs. In doing so, they will be better positioned to weather the changes that will undoubtedly come as the industry continues to contract.

What do you think?