Managing expectations
summary
Datadog migrates its infrastructure to Kubernetes starting 2018 and builds a new Deployment Platform that must absorb 2000+ engineers across 250+ teams making tens of thousands of changes per week to 3000+ workloads. An early region rollout breaks trust: technical pieces mostly work but rollout pacing and inconsistent expectations cause resistance. Recovery happens by partnering with willing teams, integrating their tools (Cadence rollout, web availability fixes), starting small, showing wins, listening, and granting ownership so the platform converges and adoption grows.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:31Infrastructure scope evolution
- 8:36Product vs platform disconnect
- 13:18Trust during rollout
- 17:11Rebuilding trust
- 20:50Human reactions and habits
- 22:53Collaboration framework
- 24:20Final thoughts
related talks
32mHow we did Platform Engineering at Datadog
51mHow to Successfully Manage a Technology Shift at a Global Scale
18mScaling DevOps in Complex Environments: Challenges, Failures, and Successes
43mTo make mistakes is human. To auto-deploy it is DevOps.
52mDevops @Scale
24m"Misadventures in Continuous Delivery"
45m200 Teams, 1 Goal: DevOps Maturity at Scale
35mPanel: Navigating the Path to Platform Engineering...