Slashdot Mirror
Michigan Sues HP Over Decade Long, $49 Million Incomplete Project
itwbennett writes: On Friday, embattled HP was hit with a new lawsuit filed by the state of Michigan over a 10-year-old, $49 million project that called for HP to replace a legacy mainframe-based system built in the 1960s. Through the suit filed in Kent County Circuit Court, the state seeks $11 million in damages along with attorney's fees and the funds needed to rebid and re-procure the contract.
I'm sure that the salesmen have long-since cashed any bonuses they received from landing the contract and moved on to other companies.
Do not look into laser with remaining eye.
Well, I can add that to the list of things I used to wonder about and wish I never found out.
At 50 million bucks, why didn't they emulate the old machinery or port the code to an interpreter running on a modern system? Off the top of my head, that sounds like the most reliable ways to duplicate exactly an old system.
That's a great question, and the answer is, IBM Z-series business unit has, bar none, the most aggressive, talented and ruthless customer retention team in the world. You're right, there's no sane reason why a mainframe application can't be emulated at least for a stopgap measure. But you'll find that there are a score of legal, political and business reasons why you won't be allowed to do that.
Moreover, you'll find that it's just impractical to port the application to any other reasonable platform. Even though your smartphone probably has more guts than the '60's era mainframe you're trying to get off of, actually making the cutover is very VERY difficult, for a variety of reasons, few of them technical.
Oliver's law of assumed responsibility: If you're seen fixing it, you will be blamed for breaking it.
True, but HP should have known this based on experience with other mainframe projects or via research on similar projects by other companies.
(1) The contract was made by EDS. HP had nothing to do with it, other than having acquired EDS.
(2) The migration is not just off the mainframe (a VMS system), it's onto a web-based platform instead, so they can get rid of both the mainframe, and the extra VT320 emulator they have to run to talk to the thing.
(3) Getting the same functionality and security of of a non-VMS system is a rather difficult endeavor, even if you use FLASK Linux or a similar purportedly secure computing platform, and add a bunch of them together and try to pretend "it's the same as a mainframe". Of all systems one can get off easily, VMS is not one of them, since it's so much better designed than most modern systems.
Scope of the task is *LARGE*
It's a doable proposition, but it would likely take (expensive to hire) 40+ year olds with experience in both sets of technology, along with people capable of parsing "business rules" out of languages like COBOL, FORTRAN, BLISS, and VAX (or DEC Alpha) assembly language, and whatever the heck else it was coded in at the time it was first deployed (depends on what they meant by "aging mainframe" in 2005).
These people would also have to be either very sophisticated in working over a "Chinese Wall" arrangement with another group doing the new systems development (not a development model most younger coders are familiar with, since you mention "interface contracts" and "unit testing" and "branch path analysis" to most of them, and they blink at you as if you've just taken a polyjuice potion and turned into Mad Eye Moody). Alternately, these old farts would need to have *also* kept current on new technology to allow them to be able to do both sides of the task.
So, you are talking expensive people in their mid 40's to mid 50's to get the job done.
Guess who were the first people let go or offered early retirement packages, to improve the profit-per-employee ratio for EDS to get the highest valuation in the acquisition by HP? Guess who were let go or offered early retirement as "cost reduction" measures in the four or five rounds of that HP has gone through since then?
It's a doable job, but I don't know of a company in the EDS (HPE now) or IBM Global Services space right now that wouldn't just start over an "fix business rules problems as they come up", rather than providing an equivalent (but now web based) system. I don't know experienced people in either of those two, since they've jettisoned all their expensive (talented) old people and replaced them with cheap (untalented) recent grads or offshoring.
If you think that's an unfair comparison on talent ... if you were a talented college grad, would *you* go to work at a company which is in the throes of a 30,000+ person layoff (something IBM did earlier this year, BTW: HP is a "late bloomer"), and in the process of spinning out the division you'd be working in? Or would you take that offer from Uber/Facebook/Twitter/Google/[anyone but IBM or HP] instead?
They are likely going to have to hire someone and PM it themselves. States are notoriously bad at that (and at spending money on their own people, as opposed to being willing to spend a lot of money on a contractor company) -- look at how Oregon and Oracle are arguing about the [still] nonfunctional Oregon State Healthcare Exchange to see what comes of hiring your own [unqualified] PM and "doing it yourself".
My cousin, Mark, could do it. Sadly, he is disabled now.
I could do it; so could a dozen or so people I could name off the top of my head (Wes Peters, for one). Sadly, we are all sane now.
They are pretty screwed; they are going to have to do a "second system syndrome" version of things, or settle with HP/HPE and pa