The Mainframe: A Phoenix From The Flames
by Robin Bloor, Partner
When the Phoenix of Greek mythology felt that its death was approaching, it would build its own funeral pyre and set it alight. Then it flew into the flames and was consumed by them. But as it was reduced to ashes, miraculously, a newly born phoenix sprang forth from the pyre. Well, the IBM mainframe pulled off a similar trick. As the old system 390 died, the newly born zSeries rose from its ashes.
The Mainframe on the Pyre
The mainframe?s approaching death became apparent in the late 1980s. IBM was attempting to lower the mainframe entry point and win business among mid-tier companies, but it was getting no traction. Instead minicomputers, some of which later morphed to become server computers, got nearly all the business and some low-end mainframe users even began to peel away. By the time the PC had a GUI, the mainframe market had become static. Suddenly Unix was the only game in town.
It was around this time that the seeds of the zSeries were sewn. IBM had become a horribly siloed company. There was the mainframe silo, the AS/400 silo, the newly born Unix silo, and the Intel based PC silo, which was endowed with both PCs and servers. IBM embarked on a long overdue program that would unite the hardware technology that lay beneath its four technology platforms.
The mainframe died in the flames of Unix, with Microsoft adding fuel to the fire as Windows NT gradually became a respectable choice as a server OS. Software packages migrated away as did customers. The funeral pyre burned brightly for about 10 years.
You could divide the mainframe market between mainframe sites where there was still a substantial portfolio of applications after you moved what could be moved easily ? and those where there was not. This latter portion of mainframe users were consumed in the flames. And yet when they were gone, there were still 10,000 mainframe customers left ? and many of these could number themselves among the 10,000 largest organizations in the world.
As the mainframe software market contracted the mainframe ISVs indulged in an orgy of acquisitions. So when the smoke blew away there were just a handful of the original players left. They included IBM itself, CA, BMC, Compuware and ASG ? and their software spanned a multitude of platforms. All of them, including IBM, are more or less platform agnostic.
IBM invested in the newly born Phoenix, the zSeries. It continued to invest hundreds of $millions each year in zSeries technology, while driving as fast as it could towards technology consolidation. It invested tens of $millions each year in education to ensure that zSeries skills were not too difficult to find or acquire. It invested tens of $millions to extol the zSeries? virtues and drive customer demand. And additionally it made two significant software investments: it gave the zSeries a very functional Java environment and it added zSeries support for Linux.
The profile of the zSeries is that of a versatile high-end server. It offers just about the highest level of single-system availability there is. It scales to the highest imaginable level and happily does so with a mixed workload. It is secure and its resources can be dynamically partitioned to deliver a level of efficiency that no other platform offers.
IBM first began reporting zSeries revenue growth in 2001 and this has continued ever since. In 2005, IBM reported that the zSeries? share of the market for servers above $250,000 had increased by 4% and its share of the above $500,000 server market had increased by over 10%.
More to the point, the percentage of zSeries spend going to supporting traditional mainframe applications has been gradually diminishing with each quarter. By the fourth quarter 2005, over 60 percent of zSeries revenue was driven by new workloads. Very prominent in this 60 percent are applications running on Linux. Linux has an extensive portfolio of server packages and nearly all are zSeries-ready.
The zSeries is particularly attractive as a Linux platform. Linux runs in a virtual partition on the zSeries and thus implementing a new instance of Linux can be achieved in minutes rather than the hours or days it is likely to take if done via other means. Additionally the just-created Linux instance is consolidated and managed from go.
The zSeries is also attractive as a fundamental part of a SOA strategy. One of the big challenges of SOA is to be able to truly manage end-to-end service levels. It just so happens that service level management is more advanced on the zSeries.
Currently, the zEcosystem generates somewhere between $4 billion to $5 billion each year. While this is less than IBM used to glean just from mainframe DASD sales in years gone by, it is still a sizeable revenue stream and it?s growing.