What is the best way to protect your investments in software/development?

Question

What strategy will you choose, to protect your investment in software (or which stragegy does your company have choosen)?

If you search strategies at Google you might find a lot of results, showing company ads without any consistent message. It seems all to be fluff.

I've heard a lot about strategic actions to improve the overall ROI. They are covering things from the deep development side, up to strategic management processes. Things from Unit Testing, Automated unit testing, Test Driven Design & Development, Static code analysis, Agile development and Continuous integration to ensure quality of your software to agile project management methods such as SCRUM or SOA(buzz).

I think it's much about interfaces like Web-Services. If Cobol based mainframe software has capabilities to export their logic in a secure way by webservices, why should somebody migrate to another platform? If somebody like to extend it, he can write new services in a more up-to-date language and chain it with the existing processes. Thats a big idea behind Service Oriented Architectures / Business process Management solutions I think. (I like P2P and Hub and Spoke all the way.)

Another big thing is the support of a big vendor like IBM. But service like this isn't cheap and might not arrive at SMEs.

How can you apply things like this in "the real world"? SOFEA/SOUI brings nice strategic improvements. I've written my homepage in JavaScript all the way (I should shame on me).

Can a whole IT-Environment without open standards be aligned into a big service oriented structure (Like Exchange/MS Office etc.)? I don't have seen a real business ready, whole IT covering structure ever.

What are other issues to protect investements you've done/you're going to do, to acquire new assets?

Answer 1

By "protect investments" do you mean not having to upgrade existing legacy systems and yet add new value by bolting new technologies onto them in a non-intrusive way?

If so, this is pretty well understood, as follows:

Either the legacy system supports a standard way of being called (or making calls) to other systems, in which case there is no problem.

Or it doesn't, in which case the solution is completely different depending on the system, so very little general advice can be given (besides RTFM).

There are tricks you can use for systems that are at first sight totally closed to extensions. For example, if a mainframe application is able to print a plain-text report, you can redirect that report to a program that will parse the report back into reusable data, and so you've turned a "print" capability into an asynchronous RPC.