DevOps: Supporting aircraft from hanger to take-off
summary
A major European airline moves from siloed Dev and Ops into product teams that release on demand, migrating almost entirely to cloud during the pandemic and managing ~1,700 repos as infrastructure as code with GitOps with Terraform and Atlantis, replacing three Jenkins instances with templated, config-as-code pipelines and staged dev/prod sandboxes. Critical systems include an electronic flight bag fuel service (wrong values can change flight behaviour); failures include a mainframe–AWS outage caused by a tunnel dig and earlier accidental volume deletions, driving a strict policy that click-ops is never okay.
AI-written summary · about
chapters (13)
- 0:00Introduction and my failures
- 3:58DevOps and aviation safety
- 5:42Legacy systems in airlines
- 7:06Mainframe team as DevOps pioneers
- 9:12Aviation technology challenges
- 11:41Making change in legacy environments
- 13:46Practical change management example
- 15:55Pandemic cloud migration opportunity
- 17:19Why change is so difficult
- 20:28ClickOps never okay in production
- 23:59Infrastructure as code at scale
- 25:24Reflecting on five years of failure
- 28:54Three lessons from failure
related talks
31mDevOps in Infrastructure — When It's Mostly Ops...
12mAir France–KLM’s GitOps Takeoff: Real Stories From the Flig...
43mDevOps@bol.com a great journey towards autonomy
36mDevOps in Outsource… Really!
27mOur DevOps Journey So Far (and Terraforming)
48mDevops@Scale
55mDevOps @Scale (Greek Tragedy in 3 Acts)
29mWe Have Always Done It This Way! Now Let’s Try Something Completely Different