What is happening to mainframe SOA?

SOA is popular everywhere, but mainframe shops would seem to be one of the most fertile areas for SOA adoption due to the potential to open up the value of mainframe investments to a wider audience, and truly leverage it.

But one issue faced by all mainframe organizations is how to get access to the mainframe applications and data sources, and make them available as SOA services.

From the extent of activity in the marketplace at the moment, it seems like others also see the potential here. A year ago, there were a number of companies playing in the mainframe integration space. Then NEON Systems was acquired by Progress Software, and more recently Seagull was bought by Rocket Software. At one fell swoop, two of the biggest players have therefore been snapped up – so what has been the effect on the mainframe integration market?

Looking at the NEON acquisition first, Progress has placed it in its DataDirect division, apparently indicating that it is the data access elements of NEON that are of most interest to it. It could have chosen to put NEON with the Sonic ESB business, positioning it more as an SOA integration tool for the mainframe, but it did not. Indeed, the announcement made it clear that Progress was trying to create ‘the unparalleled data connectivity leader’. This seems to make it pretty clear that the software will be more and more focused on purely data solutions, making it less and less suitable for general SOA needs.

As a company focused on OEM channels, Rocket is presumably going to try to hook the Seagull software into other SOA providers. Given Rocket’s history and the Seagull strengths, this will probably become a play to hook Seagull’s iSeries (AS/400) stuff into IBM – this would probably make sense to both companies, since support for iSeries is limited across the rest of the mainframe integration market. But it seems clear that Rocket’s focus will definitely not be on direct sales of Seagull’s SOA tools.

So where does that leave mainframe companies looking for integration solutions to bring the mainframe apps into SOA? One option of course is IBM – but in the case of CICS for instance, the real smart stuff is only available in IBM’s latest version, and many customers cannot move to new CICS versions with any great speed. Then there is the question of IMS connectivity as well as DB2 and of course batch.

It seems pretty clear that in the long run, IBM will provide comprehensive, easy-to-use answers to all SOA mainframe integration needs – after all, the company is totally dedicated to SOA. But for all those customers who cannot get to the new versions fast enough and yet who don’t want to be left behind in the great SOA rush, the third party choices have just become a lot more limited. Luckily, there are still some attractive and viable options in the market to choose from – at least for now.


Posted in Imported, SOA.