High Performance Backstage for High Performing Organizations
summary
Spotify runs Backstage for 50,000 components and 100,000 APIs with a monorepo hosting ~200 custom plugins, and moves from polling to a push-based ingestion using GitHub events to cut DB ops from ~9,000 to ~1,000/min and CPU by ~90%. Autonomy comes from the new frontend/backend plugin systems and using a Yarn plugin to align dependency versions; resilience comes from a plugin boot strategy plus horizontally scaled backend deployments to isolate failures and avoid whole-instance outages.
AI-written summary · about
chapters (8)
- 0:00Backstage at Spotify scaling
- 1:46Backstage as knowledge pane
- 3:51Backstage challenges
- 5:13High-performing Backstage focus
- 6:59Autonomy with new systems
- 10:35Resilient Backstage deployments
- 13:43Catalog ingestion scaling
- 17:56Recap and closing
related talks
24mScaling the Stage: Managing 200+ Plugins & Thousands of Engineers
27mShifting Spotify Engineering from Spreadsheets to Backstage
20mSpotify's Growth Exposed the Abstraction Paradox
23mScaling the Sound: Fleet Management at Spotify
27mBalancing autonomy, technical alignment, & speed through strategic collaboration
11mBackstage Plugins Made Simple: No Rebuild Needed
29mPlatformizing features at Spotify
26mDivide & Collaborate: Creating Scalable and Healthy Backstage Ba...