Is there such a thing as a mainframe monopoly?
15 October 2009
As discussed in detail in The New York Times, The Justice Department is starting a preliminary antitrust inquiry of IBM. I can’t speak to whether certain actions taken by IBM or other companies were lawful or not, but I do think the investigation speaks more to current problems in the industry than it does to any particular wrongdoing. Much of the action seems to be driven by the fact that IBM has a near-mainframe monopoly and that businesses rely on mission-critical code that can only run on these mainframes. The issue here is not just IBM’s mainframe monopoly but, rather, the fact that companies are relying on code that was written twenty or thirty years ago. I say the mainframe monopoly is not the main issue because calling it a monopoly ignores the realities of modern computing. Customers clearly have options beyond the mainframe; most (if not all) consider and purchase modern servers for production systems that either run alongside mainframes or have replaced mainframes. A mainframe monopoly is like a train monopoly… It might be a point of contention if one company owned all the trains, but this company would still be competing against all the very prevalent and modern options (trucks, boats, cars, planes) for shipping and travel. One cited “proof point” of this monopoly is that modern server prices have fallen by over 40 percent since 2001 while mainframe prices have fallen less than 13 percent. But the server costs are only one element of the total cost of ownership. The mainframe has a reputation for reliability, scalability, and throughput. At an enterprise level, servers are typically positioned as a mainframe replacement in part because of the ability to cheaply buy enough to duplicate mainframe values in volume. The total cost of ownership (which includes hardware, software, and services) for all those servers compared to a single mainframe is a point of much debate. Some state that the total cost of ownership for the mainframe is lower while server proponents claim the opposite. But if a single mainframe can potentially offer a lower total cost of ownership than many servers, it makes sense that the hardware price would remain higher. And regardless of whose numbers you believe, the modern server is positioned as a commodity compared to the mainframe; not because of the lack of monopoly status but simply because one of the values is its cheapness. If part of the server philosophy is the cheapness of the hardware, it makes sense that the price would decline more rapidly. The real issue cited in the action is not the hardware monopoly but the fact that many companies have mission-critical systems running on these mainframes; systems they have invested too much money in to move. Any company that cannot ever move off of its existing base of code because it has invested "too much" in it will one day face extinction, if it does not face it already. From where I sit, the software and services industry is fueled by the multi-billion dollar business of selling companies modern solutions and consulting to migrate off of legacy solutions. In fact, IBM, as a major force in professional services, is one of the companies helping consumers move off of legacy code to modern enterprise software. Has IBM prevented other vendors from creating options that would allow companies to retain their legacy code but run it on cheaper, modern servers? I am not going to comment on the truth or legality of such actions. My concern is and always has been about how to plan for the future IT direction of a company. Even if such legacy code alternatives have been prevented, these would only have been temporary measures anyway. If a company takes 30-year old legacy code and moves it from a mainframe to a server, that company’s problem is not solved. It may mean some lower-cost maintenance for the next few years (though, as discussed earlier, the total cost of ownership is in question), but that company still needs to consider its true next step. Otherwise this is sweeping a long-term problem under the rug with a short-term fix.