The trouble with some legacy applications is that businesses dare not get rid of them. That is fine as long as the platforms on which they run continue to be at least supported - and ideally updated - by the manufacturers.

For many reasons, that cannot always be the case, which can leave users with a problem: can they find an alternative platform, or will a development team somewhere - either in-house or at the original vendor's - be obliged to re-engineer the application?

Sometimes the user has some luck running in their favour, such as has happened in the case of Paymaster, aka the old Office of the Paymaster General. It had an important legacy application trapped on a soon-to-be-unsupported machine, but it was at least written in Cobol.

This has meant that a migration option was available to it, together with the possibilities of enhancing the application's capabilities. It does, however, raise the issue of whether the platform vendors are now exploiting technologies such as migration and code translation tools to the best advantage of long-established users, or just pursuing their own short-term self-interest.

Paymaster's problem was that its core banking suite, fairly defined as "mission critical", ran under VME on a Toshiba Trimetra mainframe, for which the vendor had announced it would cease support as of last April. The suite was the last remaining application on the mainframe and it is unsurprising that it felt running such an important system on an unsupported platform was not a long-term option.

Because it is Cobol-based, however, it did mean that Paymaster could try the migration and modernisation tools suite produced by Micro Focus. In practice, the work was carried out by Transoft, which is a member of Micro Focus' Migration and Transformation Consortium (MTC), which completed the task in some 10 months. The migration target was a Windows server platform, and the advantages to Paymaster can be seen in both direct cost savings and additional productivity.

The cost savings, estimated to be around £300,000 a year, come from not only the lower hardware purchase and support costs, but also from the fact that by porting to the Windows-based Micro Focus Cobol it meant that the existing applications code and data files could be re-used. The main productivity gain is the ability to run the banking processes up to 30-times faster, while gaining extra security capabilities.

This migration does raise wider issues, however, particularly when it comes to migrating established mainframe applications. The ongoing legal spat between IBM and Platform Solutions Inc (PSI) highlights the way in which the former, once it had lost interest in using the latter's technology to port mainframe applications and the z/OS operating system to Itanium-based platforms, decided that PSI should not be allowed to offer the technology to anyone else that might want to help existing users with such a migration.

It is, of course, fair to say that z/OS and the zSeries platform are still very much a current environment, but the fact that other hardware solutions are available as an alternative, and available from a wider range of vendors than just IBM, means there is scope for users to get similar benefits to Paymaster in terms of lower platform purchase and lifetime ownership costs - if only by virtue of the usual rules of competition in the marketplace.

It also demonstrates the point that such hardware - not just the Itanium but x86-based servers as well - have at last got to the point where they can offer comparable, and often better, performance to mainframe technologies.

Last, but by no means least, the ability to run such operating systems and applications on non-proprietary servers not only offers existing users more flexibility and choice, but also creates the opportunity for those users who might never think "mainframe solution" for their IT requirements to consider exactly that.

The upshot could be that by welcoming applications migration more applications and O/S sales would come the way of vendors like IBM than was ever possible with a policy built around the notion of a total lock in. ®