There was a time when it looked like the mainframe would go the way of the horse and buggy. I remember a meeting almost two decades ago when a senior IBM executive defended the mainframe by pointing out that dinosaurs were remarkably durable. They roamed the Earth for over a hundred million years. What was it about the mainframe that had so many of us convinced that it would become extinct? There were three key factors; the cost of the platform, the complexity of the development environment and the inflexibility of the platform to change as the IT environment changed. Had IBM not made some significant changes in the platform, the mainframe would have died.
Rescuing the mainframe
IBM decided that the platform was worth saving and continued to invest in it. The plan to save the mainframe did not emerge over night—it evolved. Leveraging the activities of a Linux skunk works in IBM's German Labs, IBM ported the Linux operating system to the newly anointed Series Z. Initially this caused some confusion among customers—but it was also seen as a sign that IBM was planning to bring its work horse into the modern era. Reducing the entry price for the mainframe was a very significant move. Soon it became possible to buy a mainframe for about $100,000—a far cry from the multi-million dollar starting price tag that the mainframe had once boasted.
But these two moves would not have been enough on their own. With only these two changes, companies would have purchased a less expensive platform but would still have been faced with an inflexible computing platform. The next two changes IBM made to the zSeries were the most important. First, IBM implemented its full middleware stack on the zSeries. Secondly, it introduced two types of processors that balanced zSeries workloads. One of these processors, called the zAAP (zSeries Application Assist Processor) enabled the zSeries to execute Java workloads and integrate web-based applications on top of the z/OS. The other processor, called the System z9 Integrated Information Processor (zIIP), was a specialty engine that was built to run database workloads in support of business intelligence, ERP, and CRM applications on the mainframe. This encouraged the centralization of data on the mainframe, pulling it from distributed computers running some of these applications.
The skills issue
While all of this is important, IBM's biggest challenge with the mainframe over the past 15 years or so has been the lack of zSeries skills. IBM is trying to solve this problem by partnering with some universities to help improve the number of students who might take an interest in the aging platform. While there is still a way to go before the need for skills is addressed, there are a number of educational partnerships in place, with more promised. IBM has the goal of educating 20,000 students on the mainframe by the year 2010. In addition, the company is revving up its partnerships on the System z. Today there are 1500 business partners and 1350 ISVs that port their software on the zSeries and currently 30% of zSeries revenue is driven through partners.
Ironically, IBM's biggest challenge comes from the ingrained beliefs of customers. Many take it as read that in order to move forward with their computing infrastructures they have to decommission their mainframes. Many of those I speak with are not aware of the changes IBM has made in the mainframe and they continue their gradual migration from the platform. Ingrained beliefs are hard to change when it has been "conventional wisdom" for more than a decade. At the same time, the message seems to be getting through in some places if the improving sales of the mainframe are anything to go by. Indeed, some relatively small businesses are even adopting the zSeries—an unimaginable prospect just a few years ago.
Reliability, Security and Manageability
Can the improved position of the mainframe be completely attributed to the openness of the platform and the additional students lining up to study its intricacies? While these changes have transformed the dinosaur platform into a fairly hot platform, I think the reason for the zSeries resurgence has more to do with reliability, security and manageability than anything else. Companies trust the zSeries to run and manage their mission critical transaction engines. Security based on the tried and true RACF model is less risky than the security environments on any other platform. Such things are increasingly important as data centers continue to expand in size and thus the strengths of the zSeries become increasingly prominent. Maybe that IBM executive was right about the life expectancy of this particular dinosaur.