Surviving Long Event History

on YouTube ↗38m

" summary

NRK runs event streams 13–15 years long and replaying them is practical if events are versioned, small and schema-based. Uses Protobuf over JSON for compact, stable schemas; avoids polymorphic serialization and giant payloads (subtitles should be file references, not inline). Applies a hybrid snapshots + replay strategy for performance, stores outgoing messages with outbox pattern, and runs annual in-store migrations to prune obsolete origins and fix corrupted events.

AI-written summary · about

" chapters (17)

1. 0:00🚀 Introduction: The Importance of Long Event History in Event Sourcing
2. 2:00Event Sourcing and Why It Works for Long History Survival
3. 4:00Key Benefits of Long Event History and Why You Should Keep Events Long
4. 6:00Common Challenges: Performance, Storage, and GDPR Concerns
5. 8:00The Hybrid Approach: Combining Event Replay with Snapshots for Performance
6. 10:00Best Practices for Event Versioning and Data Ownership
7. 12:00Separating Commands, Persistent, and Transit Events
8. 14:00Avoiding Overgeneralized Event Types: Why Specificity Matters
9. 16:00The Dangers of Large Event Payloads and How to Handle Them
10. 18:00Event Serialization: Choosing the Right Format for Long-Term Survival
11. 20:00Migrating Events: How to Prepare for Event Migration
12. 22:00Protobuf vs. JSON: Why Protobuf Is Better for Long-Term Systems
13. 24:00Handling Broken Backward Compatibility and Event Cleanup
14. 26:00Using Event Versioning vs. Creating New Events
15. 28:00How to Manage Event Migrations in Real-World Scenarios
16. 31:00Key Takeaways for Ensuring Long Event History Survival
17. 34:00Closing Thoughts: Event Sourcing and Building Systems for Decades

" related talks

• 01h 03mSurviving Long Event HistoryVagif Abilov · NDC London 2025
• 43mEvent Sourcing dealing with an eventful pastHannes Lowette · Build Stuff 2025
• 49mEvents... And what next?Paweł Świątkowski · GrillRB 2019
• 51mThings I wish I knew before I started with event sourcingMichał Ostruszka · Confitura 2023
• 42mSlow Event Sourcing reprojections? Just make them faster!Dennis Doomen · KanDDDinsky 2022
• 52mUse the Source, Luke: High-fidelity history with event-sourced dataKeith Gaddis · RailsConf 2012
• 48mEvent Sourcing - what could possibly go wrong?Andrzej Ludwikowski · Jfokus 2022
• 51mHow do we approach event streaming at Revolut?Jacek Spólnik · GeeCON 2019