The ROI of Developing Applications with CA Gen

Here is a whitepaper on CA’s site. It details how CA Gen-generated applications have achieved massive Return Of Investment (ROI), and how that ROI can be extended and maximised.

I agree with pretty much all it says in the post, but what it doesn’t say is that ROI can be maximised only if you’re astute enough to design your applications with future platforms and technology waves in mind, such that they don’t need changing when those platforms come out. The ROI can be FULLY maximised if you’ve done that, but only ALMOST maximised if you haven’t.

I suppose what the point here is that whereas an application component (or complete application) will return the maximum ROI if there are absolutely no changes to be carried out when technology paradigms change (such as batch to online, or client/server to web), or when a platform upgrade is required, such as a mainframe to a distributed system.

This leads to the point that when any change is being considered, the benefits that Gen can give are reduced slightly – NOT, it has to be said, by a failing in Gen! – No – those benefits are reduced by the very act of changing paradigm.

This, of course has to be tempered with the big ROI that Gen has in the first place – quoting from the paper itself:

“When compared with hand-writing COBOL, CA Gen takes 40 percent of the time to develop any application.

And when compared with Java IDE’s, CA Gen J2EE development takes just 60 percent of the time.

Compared to IBM/Rational development tools, CA Gen is more robust, reliable and reduces overall project costs.”

Mr. Pipalwa COO, Stringray Technologies

Therefore, the big “hit” in terms of ROI with Gen, you’ll get at the start of an application’s life (and continue to reap the rewards as the application evolves, due to the similar productvity gains). But, that will be eroded by a technology change, ONLY WHEN TECHNICAL CHANGES ARE REQUIRED TO SUPPORT THE NEW TECHNOLOGY OR PARADIGM. When changes aren’t required, then that ROI is completely maximised.

I suppose that I got to thinking about this sort of stuff because I am aware that there are lots of “upgrading” and “platform change” type projects going on in the Gen world right now – I suppose that those people that have taken part in these sort of projects have done so, only after considering the advantages and disadvantages of a complete rewrite versus reusing the Gen components (see here and here).

Wrapping up, then, I think that if someone were to construct a graph of a piece of code’s value, plotted against time, compared with other languages (COBOL, Java, as stated above), the value of a Gen action block would go down slower than that of a COBOL module over time, due to the lower initial cost. But, come a change in technology, the COBOL module would be thrown away, resulting in zero value, whereas the Gen action block would increase vastly – 100% in fact - since it wouldn’t need to be rewritten – even if it did, it would still be worth (say) 75% more than before a technology change!

Does this mean that the value of a Gen component actually grows over time ??????