The Resurgent Mainframe: A Platform for Innovation
Published: August 19, 2008
by Chris Craddock
Innovation may not be the first thing you think of when someone mentions mainframes. After all, the mainframe has been pilloried in the press and on Wall Street for much of the last 20 years. Some of that criticism stemmed from a bias toward newer platforms, but much of it was warranted, prompting IBM and independent software vendors to deliver better technology and to make mainframes easier and more cost-effective to use.
Since the mainframe's debut in April 1964, IBM has been constantly reinventing it to keep pace with the evolving needs of enterprise customers. The technology has progressed from punch cards and batch to online banking over the public Internet. Today's z10 mainframe systems are radically different from their ancestors. They are much smaller, faster, cheaper, and more energy efficient, so there has undeniably been a great deal of technology innovation over the last four decades. And that innovation continues today.
Amid all of that change, however, there has been a constant focus from vendors to protect the value of the customer's technology investment and to preserve the core architectural values on which the platform is based. Business applications written decades ago still run on a System z today, and the terms often associated with mainframes are comforting words like "mature" and "reliable." These mainframe characteristics are all good, but they hardly get young people's attention or generate a buzz on Wall Street.
Although the industry rarely received much credit for continuous innovation in mainframe environments, that's beginning to change for the better as customers continue to grapple with the growing chaos among their distributed systems. Moreover, since customers vote with their wallets, they're much less influenced by the press than by their own internal business needs. Successful businesses balance their technology needs against fiscal constraints and risk management, so while distributed systems are universal today and provide important business functions, the mainframe continues to be the workhorse of core business processing for most large enterprise customers.
It's easy to see that the way customers use their mainframe systems is very effective at generating slow, steady growth and profitability for the vendors that support it. But many deep thinkers in our industry are struggling with the question, "Is it possible to have a growth business in the mainframe space?" It is, but there are three key areas where innovation is needed.
Energy and Floor Space
Data centers used to be occupied by locomotive-sized mainframe systems and their equally massive disk subsystems. With the introduction of CMOS processors and commodity disk subsystems in the mid-1990s, the mainframe footprint shrank dramatically, but around the same time the number of non-mainframe systems exploded, so now our data centers are packed to the rafters again and the utility bills are skyrocketing.
Power and cooling are issues for everyone but never more so than in the distributed systems world. One of the most cost-effective ways out of that thicket is to leverage z/VM and zLinux to consolidate some of the server sprawl, at least for the Unix and Linux systems out there. The overall energy efficiency of the mainframe is already compelling and that should lead to a growth spurt for the mainframe as a technology platform. This trend will drive the need for management tools to make it easier to provision and manage very large numbers of virtual servers. Once the distributed work resides on the same physical processor as traditional z/OS work, new opportunities will accrue to both mainframe and distributed applications. Distributed work can more easily leverage secure z/OS database and transaction processing capabilities and the z/OS work can easily leverage distributed processing capabilities. There will be business value to both sides, as well as to the bottom line.
One of the key differences between the mainframe and the distributed systems world is that there are many fewer moving parts on the mainframe. In the distributed world, there are many separate components to piece together and their implementation and ongoing management can be daunting. In the mainframe world, there is also a lot of complexity, but it is on the inside. A relatively small number of people can successfully manage a mainframe sysplex cluster, but only if you're willing to settle for a minimum of customization.
Anyone who has read through the blizzard of implementation guides and configuration guides for the z/OS operating system and the major middleware subsystems will be painfully aware that there are literally thousands of knobs that can be twiddled. Many of them require customer specification because their default choices are barely functional, so getting the best business ROI from the platform can take a great deal of effort. That configuration complexity has become a barrier to success, particularly as the old hands head toward retirement. One key innovation would be to provide software solutions to mask or eliminate much of that complexity. After a great deal of customer pressure, IBM itself is now promoting a theme of configuration simplification, so this is an area where IBM and the independent software vendors could compete and assist each other at the same time.
User Interface and Usability Improvement
Jim Porell, chief architect for System z software at IBM, has famously said that the mainframe is the only major technology platform where the on-ramp to the information super highway is a rutted dirt road. As an industry, we got away with neglecting the user interface because our user community was a bunch of smart people who had spent most of their adult lives working with TSO and ISPF, or CICS BMS or IMS MFS. If the interface was hard to write, it ought to be hard to use, right? We old hands tend to regard mastery of the mainframe's multitude of unfriendly interfaces as a badge of honor.
However, the next generation of mainframe administrators will begin to take over the reins soon and if they have to use 3270 terminal emulators, things are going to get ugly for the GUI generation. Ironically, there is no technical reason we have to put up with poor user interfaces. The mainframe already supports all of the Web services and GUI technologies that would enable the mainframe to fit seamlessly into the rest of the customer's network. We just haven't made the mainframe a comfortable place for the next generation to work. At the same time, we have been passing up a great opportunity to make life easier on ourselves by making the mainframe a first-class citizen again. As an industry, we can and must do a lot better.
If the industry took up the challenge of innovating in these three areas, then I think there is a very real chance we can actually deliver growth while still preserving the core platform values that customers have relied on for more than a generation. That would allow the Gen Y kids to work comfortably and effectively on the descendents of the same systems their parents and grandparents did. Wouldn't that be a hoot?
Chris Craddock is a 30-year industry veteran with 20 years of commercial product development experience. He is currently senior vice president and distinguished engineer in the Office of the CTO at mainframe software maker CA.
The Modern Mainframe: A Model of Space and Energy Efficiency
A Mainframe Renaissance
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot