The fundamental question is straight forward: is there growing a new market potential for mainframe operating systems and applications software outside of their traditional box'? And if there is what is the best way to exploit the opportunity?
If you look at the evidence growing in the areas where that market potential might exist, there are strong signs that existing solutions are about to start running out of steam or, more to the point, prove themselves unable to exploit the ‘steam’ being generated.
It can be argued that even the smallest, cheapest commodity server has, if placed in any business context, got to now play its part in a complex, virtualised service delivery oriented environment. In just a couple of years such servers will in practice be running parallel processing engines—16 or even 64-core x86 processors—and if the applications are still architected as they are now there are likely to be a growing number of deeply frustrated users.
The move to multicore processors across even the smallest datacentre will force important, highly disruptive changes in application development and design—and when that happens every and any technology will merit serious consideration for a place on the table: even supposedly old technologies like the mainframe system.
Typical application areas where this change is likely to be most profound is in the management and operation of any web-based transaction—and transaction management, in its broadest sense, is at the heart of most web activity. While there will be many new operational and service delivery applications developed that can exploit the growing potential of multicore processors, there are tools available that already have the capabilities needed to manage many of the results those operation and service-based applications produce. This is particularly important where those results equate to managing revenue, costs and margins. These might not be cool functions for the cyber-warriors, but they'd soon be unemployed without them.
So, mainframe applications, particularly in the database and transaction management space, have the potential to take a more proactive role in managing online commercial services. This is particularly so now that companies like DataDirect have made the integration of the mainframe platform and the standard integration protocols of web services, such as BPEL, SQL and ODBC, significantly easier with its Shadow V7 offering. Mainframe applications can now be directly integrated into a growing range of online, event-driven, transaction-based tasks, adding their performance, reliability and superior I/O handling capabilities to allow for greater traffic throughput—which in turn equals more revenue and the potential of better margins for the business.
At present, of course, such a potential advantage is essentially restricted to those enterprises already in possession of a mainframe hardware platform. There is, however, now an arguable need for the re-emergence of the old Plug-Compatible Manufacturer (PCM) model in the marketplace, based on new technologies. That certainly seems to be an opportunity seen by NEC and HP. These companies have formed agreements to provide Platform Solutions Inc (PSI) with enterprise server technology which PSI combines with it's mainframe technology to run mainframe operating systems—and thus applications—on hardware running Intel Itanium processors.
PSI has a long, if indirect, ancestry in the mainframe PCM sector, having been started by ex-employees of PCM pioneer, Amdahl. It offers two variants: the high-end 6420 series which can have up to 32 Montecito Itanium 9000 processors, meaning 64 cores and 225 to 2,000 MIPS in 40 MIPS steps, and the low end 5120 series with 4 Montecito Itanium processors offering from 26 to 350 MIPS. The latter is also sold through T3 Technologies, which was one of the biggest suppliers of Flex-ES systems while that product was available.
As well as providing a lower-cost and more widely sourced alternative platform the PSI solution offers the advantage of running multiple platform environments in the same box. The ability to run Linux, UNIX, Windows and mainframe applications on the same box under a single management environment could prove a distinct advantage in integrating the front-line, web services delivery applications with the back-line business and transaction management capabilities of the mainframe. PSI's President and CEO, Michael Maulick, professes a strong belief that this helps reinforce the capabilities of the mainframe and can extend the revenue base for IBM software by linking the company to customers that it cannot presently serve with mainframe applications.
Maulick suggests that some 60–70% of all corporate data is still controlled by mainframe applications and, while many new enterprise applications are now coming from the open systems side the essential operational discipline and ability to execute reliably still remains with mainframe systems and applications. "It comes back to data—where is the master copy of the data and why can't a user have access to it from any platform that he wants?," he asks.
It has to be noted, of course, that IBM started legal proceedings against PSI some 15 months ago, charging it with patent infringement. PSI has counter-sued, both in the US and European Community courts, citing abuse of market dominance.