I'm truly happy with this one, folks!

A year or so ago, a very large insurance company went down the gurgler here in Australia.

Having been involved with their conversions of data from ancient, dinossaur mainframes to an Oracle/UNIX based system, I got quite pissed-off when the company that picked up the bits and pieces decided to re-convert everything back to a mainframe-based, Polisy system.

Their claim, well-publicised in the industry back then: the Oracle system was so "piddly" and "trivial" that it would literally take their "IT experts" absolute tops a couple of months to convert everything back to a "serious" system. And the extra load would barely register in their "real grunty hardware": an IBM mainframe.

After all, all those Oracle people had just been "bleeding dry" the IT department of this company.

NOW, the reality check.

13 months later:

• 20% of the original data and a very small subset of the functionality MIGHT be ready for loading into the live mainframe system next month.
• The existing live system mainframe is grinding to a halt everytime said 20% of the "piddly and trivial" amount of data is loaded and tried on-line.
• The "IT experts" are at a loss to explain in terms of "lack of tuning" and "bad capacity planning" what EXACTLY went wrong with the rosy picture painted last year. Oh, they blame a lot of other things. The usual scenario: bad design, bad developers, inefficient coding, yadda yadda.
• The people who knew exactly what would happen are now laughing their brains at the predicament of the new management, who put their balls on the table for delivery of the "new" system.

Hope they all get the sack, like they so well deserve... But knowing full well how these things work, this is what will happen:

IBM will step in, sell them an absolutely obscenely expensive upgrade and some immorally high-priced consultancy. Then when the thing is barely up, it will be proposed for some weird, wonderful and completely anonymous international IBM award.

As an example of another successful move away from the "dreadful" UNIX/Oracle combination. Never mind at what cost, that won't even be mentioned...

Another example of the "magic" of "scalability" a-la IBM mainframes. And some want them back? Yeah, right! Good luck.

Sorry for the venting. But this is one IBM will NOT get away with, without the truth coming out in the open at least once.

Details consisting of names, times, places and prices gladly available on request.

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Sat Feb 23 2002 - 05:57:23 CST