Scheduled Downtime vs. Seamless Conversion
I fly a lot. I did before I was a consultant, but as a consultant, my dependency on air service is very important. American Airlines effectively fired me as a customer when they dropped the proverbial “nerd bird” flight last year which was the Austin to San Jose non-stop flight. The last few nerd bird flights there was lots of on board discussion on Central Texas to Bay Area route options after the sudden termination notice we were given by American. Most of the folks I spoke with had 1 to 2.5 million miles flown on AA. That is a huge amount of butt time 500+ mph chairs!
When I’m on the planes I frequently work. But many times I schedule some personal downtime and have a gin-n-tonic and watch a DVD. Everyone needs scheduled downtime. Most of you Dev2Ops readers are probably burning some of your personal scheduled downtime right now reading this post. I assume Dev2Ops readers also don’t get a lot of scheduled downtime for the online business infrastructure they support (if ever!). So I was blown away when I got this email from my new transportation services vendor I hired…
First off, let me say jetBlue has a great service picking up poorly serviced travel customers like the recently terminated nerd bird flock and their prices are just so reasonable. The in flight TV is nice so you can do things like get the latest news from Haiti. JetBlue is donating their services to get personnel and supplies in which is just fantastic to see.
But now the eBusiness technologist part of my brain kicks in and says “WHAT!”. An airline business running without a ticketing system for two days intentionally. Scheduled downtime; are you crazy? I can only imagine what kind of bind jetBlue is in with their technical infrastructure. Surely the CEO signed off on such a project plan so the issue must be nasty. I’m left with so many questions. Why can’t you phase in the new ticketing system by route and gradually obsolete the older system? Why can’t you do what Facebook describes as “Dark Launches”. Why can’t the new system run 5 minutes after the old system is powered off?
I’m asking the wrong questions.
You have to take your technologist hat off and put on your business hat on and ask different questions. What is the cost difference between a seamless conversion and a scheduled outage conversion? Would jetBlue have to raise air-fares or ask the shareholders to suffer losses? Does the complexity of the requirements to implement a seamless conversion put the conversion out an extra year hindering the business? Does the added complexity of a seamless conversion add tremendous risk to business operations? Having done numerous transaction system conversions in my career, I can easily say a seamless conversion is probably 4x the size of a scheduled outage conversion (or more). Minimizing complexity substantially reduces risk by a greater than the 4x rate as the relationship to risk and complexity is non linear! Case in point, if the business singed up for the added expense, and schedule delay, what business impact would occur if the technology effort failed??? I would imagine given the above email, the risk was too great.
I’m about to head to the SFO airport and jump on a jetBlue flight who by that time won’t have a ticketing system. My travel schedule makes me personally interested in jetBlue’s success in this conversion obviously!
BTW – I choose United non-stop to San Francisco and jetBlue non-stop on the way back. Good news… Alaska Airline service picks up mid March to San Jose!