How success can take you to the brink of failure: a case study
summary
Accenture’s rapid adoption of Octopus Deploy overloads a single RavenDB-backed instance with 300+ teams and ~700–1,564 tentacles (3,000+ targets), causing daily crashes, long reindexes and a NuGet handle leak; engineers build a custom migrator, migrate teams to isolated node sets (paired Octopus instances sharing DB/file share), move packages off HTTP NuGet to file shares, and convert Raven to SQL. Result: stability, 13–14k deployments/month, sub‑minute deploys and an estimated seven‑figure time recoup.
AI-written summary · about
chapters (21)
- 0:00Intro
- 1:18OCTOPUS DEPLOY
- 1:55ACCENTURE
- 3:04DEPLOYMENT AUTOMATION
- 4:04PROBLEMS TO SOLVE
- 9:02DISTRESS SIGNAL
- 17:11BURNED BY UPGRADES
- 19:43CONSTANTLY FIGHTING FIRES
- 23:39DO NOT EXPECT PERFECT
- 27:25ADDED BONUSES
- 28:46THE PLAN
- 29:27SUGGESTED PATH
- 30:26THAT'S NOT GONNA WORK...
- 30:58RE-FACTOR
- 32:22AUTOMATE EVERYTHING
- 33:00A FEW HICCUPS
- 34:19PROCESS
- 35:326 STAGES
- 38:28COMPLETE STABILITY
- 40:46ENTERPRISE NUMBERS
- 44:18THANK YOU!
related talks
55mDevOps @Scale (Greek Tragedy in 3 Acts)
52mDevops @Scale
46m10+ Deploys Pe
20mLessons Learned from Building a Deployment Process for 100+ Devs at a Large Bank
26mNo Whiteboards Allowed
53mThe Road to Continuous Deployment: a Case Study
27mTaming the Kraken - How Operations enables developer productivity
15mOne Million Builds per Year, Only One Page - Operating Internal Services...