This research looks at the strategies of the four key mainframe vendors in Europe, the Middle East and Africa (EMEA): IBM, Fujitsu Siemens Computers (FSC), Bull and Unisys. Users should develop plans to manage their legacy portfolios and decide where to put new applications.

• The EMEA mainframe market is large and relatively stable; mainframe technology providers continue to invest in their platforms.
• Application dependency on the platform is a key aspect of migration strategies and will drive many aspects of hardware platform selection.
• Vendor statements about product viability and road maps will continue to be key parameters for user decisions about platform choice.

• Users should use IBM's weaker third quarter of 2007 mainframe results to drive higher levels of discounts in end-of-year purchases
• FSC customers should evaluate alternative platforms as part of a risk evaluation, taking into account FSC's SPARC enterprise servers (SES).
• Bull customers should be encouraged by the company's financial performance and strategic plans, and take advantage of the multiple operating system platforms to ensure a smooth transition of legacy workloads.
• Users should welcome Unisys' transition to the Intel architecture for its mainframe systems and focus attention on maximizing the benefits of metered pricing.

The mainframe market in EMEA is worth more than $2 billion a year and is showing a remarkable resistance to change. Although IBM is the largest player by far — accounting for 77% of revenue in 2006 — Bull, FSC and Unisys are strong in key geographies and in vertical industries such as financial services and transportation. During the past six years, significant growth and penetration of x86-based servers have occurred, along with a declining reduced instruction set computing (RISC) Unix installed base. However, each mainframe player has generally held its ground in sustaining investment and, in some cases, growing, which is remarkable given the changes in infrastructure topology trends. Figure 1 shows a comparison of the server market segmented by CPU technology in 2001, 2006 and the 2012 forecast, mainframes are represented in this figure by "others." Although these vendors have seen a small, but steady, flow of customers leaving the platforms during the past 10 years, there has not been the mass exodus that the industry has repeatedly predicted would occur. Through a variety of strategies, the four players have attempted to revitalize their business. Figure 2 shows the relative size of the mainframe market by operating system. Here, IBM is represented by z/OS/OS390, FSC by BS2000/OSD, Bull by GCOS and Unisys by MCP/OS2200.

IBM has a strong position in the EMEA mainframe market; it has seen significant growth in market share, revenue earned and the number of MIPS shipped. The spread of mainframe customers is comprehensive across the EMEA region, which means the overall business is not held hostage by the economics of a particular country. The EMEA region has a large number of midsize companies and IBM's z9 Business Class machine has proved to be popular with many of these customers. There has been a good uptake of Linux (using the Integrated Facility for Linux specialty engine) on the mainframe in EMEA, and customers generally seem positive about embracing new workloads. Therefore, Gartner sees good growth for the other specialty engines (zIIP and zAAP for DB2 and Java workloads, respectively). IBM has made strong efforts to build up mainframe skills in the region by working with a number of universities to put together training courses.

The third quarter results for IBM showed a 30% drop in worldwide system z revenue over the third quarter of 2006, and a commensurate 21% drop in MIPS. Although third quarter 2006 was a stronger-than-average quarter, it can be difficult to improve on because of the nature of the sales cycles — large account wins can make the difference between a good quarter and a bad quarter. For example, IBM EMEA revenue (across all product lines) grew by 11%. The mainframe business is cyclical, and this drop in revenue is more symptomatic of buying behavior than it is of any fundamental issue with IBM's mainframe products. To judge the business more fairly, revenue bust be compared year-to-year, rather than over shorter periods. However, in light of a weaker third quarter, IBM customers should take advantage of these results by being more aggressive in their negotiations in the fourth quarter of this year, because there is a push to close deals before the year-end. However, IBM will announce a next-generation mainframe in the first quarter 2008 and users should push IBM for details before making significant purchasing decisions this year. With respect to new workloads, we encourage users to take an open view and consider all platforms, whether they are RISC-based or x86-based. Users should also benchmark actual costs and technical performance to get a true comparison of running applications on different platforms, rather than using generalized figures (which can be misleading).

FSC has a scalable rang of mainframes based on the BS2000/OSD operating system. The company has a large and significant installed base, particularly in Germany, Austria, Belgium and Switzerland. The product range consists of the SX100 to SX160 business servers at the low end. These systems are based on SPARC64 processors and run, in various combinations, BS2000/OSD, OSD/XC and the Solaris O/S (not on the SX100). At the high end, the hardware moves to the CMOS processors with the S165 and S200 business server. These systems run the BS2000/OSD operating system and Linux. With this combination of platforms, FSC provides its installed base a scalable platform range, but without the options of running other platforms on the same piece of hardware.

FSC mainframe customers tend to be loyal to the platform and with all other mainframe markets, there is limited opportunity to attract new customers. Current users should consider alternative platform strategies for new applications. Although there is no urgent need to migrate the legacy applications off the mainframe systems, customers should do risk assessments of their applications' dependencies on these platforms as a contingency exercise. This review should be done annually and in close consultation with FSC to ensure that any changes in direction regarding support and future development of the BS2000 environment can be acted on quickly. For example, the updated SPARC-based platform the SES servers could provide a renewed value proposition to the PrimePower equivalents.

In fiscal 2005, Bull achieved its first year of revenue growth since 1999 (3% growth). After a few difficult years between 2001 and 2003, the company was successfully recapitalized in 2004. With a new executive team in place, the company launched Horizon, a strategic vision plan to further transform the company, grow revenue and improve profitability by targeting new market segments such as services, open source and HPC. Under strong leadership, the operational execution of business change has proved to be quite successful, with the latest results in the third quarter of 2007 showing a 10% increase on orders quarter over quarter, with revenue of 243 million euros.

Bull's mainframe-class business is centered on the General Comprehensive Operating System (GCOS) and the NovaScale product line. The hardware is Xeon- and Itanium-based, and is able to run GCOS, Windows and Linux applications. This product strategy has significantly slowed down the rate at which customers are leaving the platform, particularly outside France. Gartner receives few queries on users wishing to take an urgent migration approach. The combination of strong management, a strengthening financial position and a clearly articulated product road map has reassured large Bull mainframe users that, in the short- and mid-term (the next five years), there will be a degree of stability. Furthermore, the ability to run multiple operations systems on a single platform has created numerous technical options for users. For example, some have decided on a "milking" strategy where they will neither grow the GCOS footprint, nor actively diminish it. They will simply put all new applications on different operating system platforms. An alternative strategy employed by other users is to migrate off the legacy mainframe environment over a period of three to five years. This is achieved by migrating the applications onto Linux, which can be done on the same hardware platform. GCOS customers should develop a midterm strategy based on their application portfolio to take advantage of the multiple operating system platforms supported by the company.

Unisys's ClearPath mainframe servers have a parallel complementary metal-oxide semiconductor (CMOS) and Intel architecture road map, and are able to run the legacy MCP and OS2200 operations systems as well as Linux and Windows on the same hardware platform. The transition is happening at a good pace, starting with the low end of performance and progressing through the range. This was a necessary step to provide customers with the option of maintaining legacy applications and developing new ones on Windows and Linux without having to run two hardware architectures. The strategy has proved successful in that many customers are happy with the newer technologies from Unisys. However, Unisys has not focused attention on winning new mainframe customers (such as IBM) and will, in the long term, be faced with a diminishing installed base.

Unisys has also focused attention on developing an easy-to-use pay-per-usage for its mainframe customers. With good metering software and a number of options available, customers like the usage model.

When deciding on a long-term strategy, mainframe customers should realize that Unisys is becoming a services company and will focus management resources in this area. However, it has stated a commitment to customers "to sustain ClearPath for the next 15 years," promising product enhancement. Although the installed base will slowly diminish, the company cannot afford to disenfranchise large and important customers by neglecting the future needs of its mainframe installed base. As a result, users should annually assess the criticality of their Unisys mainframe applications and gain reassurances from the company about future road maps. At the same time, they should work on a strategy to develop new applications on nonlegacy operating systems, such as Windows and Linux.