FlyerTalk Forums - View Single Post - eUpgrades - an observation
View Single Post
Old Jun 21, 2016 | 12:20 pm
  #12  
epiphani
5 Years on Site
 
Join Date: Jan 2015
Location: OSL
Programs: AC SE
Posts: 306
Originally Posted by adam.smith
I'm sure you techxperts could do a better job, but is this really a tech issue?

I feel like ultimately this is a problem more of leadership and culture. If you tell IBM to build a single, integrated website that lets you mange PNRs, track mileage, eUp, etc all in one place, I'm sure they can do that. Someone else might do it better and/or cheaper, sure, and that would be great. But unless the people at the top are prepared to break down the silos and make some tough decisions, you could have the most brilliant people in the world doing the coding and it would still suck.
There might be some interesting technical challenges, but on the scale of "tech problems", the entire thing rates pretty low.

The leadership and culture issues are also probably pretty major - which is why, like cc said, you'd need to come in at the top.

The biggest problem, in my mind, is the existing technical debt and existing systems that work "well enough". All considered, AC's IT actually works quite well at its core business - ticketing, PNR, checking in, flight management, etc. The problem is that it's all been built over time since the 80s or even 70s, and there are huge swaths of stuff that reflected the best practices (by probably mid-level developers) of the time.

How would I handle that? Toss out the entire stack, and rewrite from scratch using modern distributed backends and tools - in parallel to the existing system, feeding from the existing system, running quiet and hidden until it is stable and correct enough that it can be slowly phased into production as a full replacement.

Multi-year R&D cost on top of the existing system with no tangible benefits for probably the first 18-24 months. But, when it's done, massive reduction in cost and massive increase in reliability, performance, and maintainability. Also massive improvement in analytical tooling, business intelligence, marketing, preventative maintenance.. probably improvements in OTP and other rather visible stuff.

Edit: Realistically, tossing the entire stack is unnecessary and probably too big a risk. But as an outsider, one can dream.

Last edited by epiphani; Jun 21, 2016 at 12:41 pm
epiphani is offline