I had a nationally recognized project manager tell me once on a huge project that I needed to take Cortez's approach. Cortez burned his ships when they landed in the new world, so there was no way to back out. He recommended a big-bang no-retreat implementation. I insisted on an incremental rollout with the legacy systems still in place.
The guy was implementing the same system I was, but from a different vendor and for a different company. His implementation failed miserably. Mine was a success: in fact it was the fasted company-wide implementation of the system ever.
If I were doing ObamaCAIR website, I would have implemented in a small state first after extensive stress testing. I would then have rolled out to possibly two additional states once the initial implementation was stabilized.
Even better, I hopefully would have turned the gig down once it became evident they weren't committed to the success of the project. (Then again, $600,000,000 goes a long way ....)
gitmo, you are so right on this
“If I were doing ObamaCAIR website, I would have implemented in a small state first after extensive stress testing. I would then have rolled out to possibly two additional states once the initial implementation was stabilized.”
managers never want to hear about the timelines associated with reasonable testing, politicians even less so
if you want a good laugh, take a look at Oct 3 “Ongoing Issues” where they list “fixes in the works for: the income issue, SSN, ”
http://www.scribd.com/doc/180609255/HealthCare-Gov-pdf
how on a project like this could anything related to Social Security Number be an issue on D Day + 3 ?
“I had a nationally recognized project manager tell me once on a huge project that I needed to take Cortez’s approach. Cortez burned his ships when they landed in the new world, so there was no way to back out. He recommended a big-bang no-retreat implementation. I insisted on an incremental rollout with the legacy systems still in place.”
I will agree with this to a point - sometimes there is an establishment in place who will maintain and milk a creaky, unmaintainable system that cannot scale until they retire or die, and they will defend it and sabotage anything new to replace it because their careers (so they think) depend on it.
However, in most cases, a flash cut with no fallback is suicide.
I face problems like this (right now, in fact) on a smaller scale every week, in a real-time, dynamic environment where sometimes changes have to be made without as much testing as I would like. But if you aren’t making mistakes, then you aren’t doing anything...the big questions are how fast can you recover from them, will the customer notice them, and what can you learn from them.