CI/CD Observability, Metrics and DORA: Shifting Left and Cleaning Up!
summary
Concrete claims and numbers: DataDog reports a test impact analysis that saves about 6,500 CI hours per month (~19 hours per developer) and exposes failures by cause, and Slack moves from a 20% to a 98% success rate after automating flaky-test isolation and Jira ticketing. DORA’s four metrics remain the north star (deployment frequency, lead time, change failure rate, time to restore). GenAI adoption correlates with delivery throughput down 1.5% and stability down 7.2% per 25% adoption; dashboards and a CI health dashboard make fixes visible and actionable.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 2:31Software delivery origins
- 7:08CI/CD ownership problem
- 11:24GenAI impact on stability
- 16:19Shifting left for observability
- 25:08Fixing flaky tests at Slack
- 28:35DataDog’s CI/CD improvements
- 32:04DORA metrics for SDLC
- 35:15Key takeaways
- 35:58Q&A
related talks
28mAccelerating Success: How to Optimize Value Delivery with DORA
22mDORA Metrics: The Key to DevOps Excellence
39mDev Team Metrics that Matter
19mAccelerating Code Quality with DORA Metrics
25mTroubleshooting at Scale: What DORA Metrics Miss
22mDev Team Metrics that Matter
37m#DataDrivenDevops
50mFast Doesn’t Mean Fragile: Delivering AI-Powered Software at Scale