EventSourcingWithEvent Store
summary
Greenfield Polaris uses EventStore (replicated 3–5 nodes) with ESJC clients in Java/Go across ~20–30 services to make events the source of truth; projections land in Postgres so a migration that used to take a weekend becomes ~10–15 minutes. Architecture uses catch-up subscriptions (replay from event 0) vs live-only subscriptions, recommends per-entity streams for GDPR and deletion, warns against derived events that duplicate on replays, and notes EventStore supports millions of streams.
AI-written summary · about
chapters (30)
- 0:00Introduction
- 1:33My Experience
- 1:59Project Overview
- 3:19Event Store
- 4:07Motivation
- 6:09Terminology
- 8:54User Object
- 9:35Writing Events
- 15:44Field Validation
- 17:09Write Confrontation
- 17:48Write Good Event
- 21:51Database Migrations
- 26:13Edge Cases
- 27:02Corrective Events
- 27:53GDPR
- 29:11Data Migration
- 30:18Tradeoffs
- 30:48Timing
- 32:31Subscriptions
- 33:33Message Queue
- 34:29Replication
- 35:20CQRS
- 38:57Summary
- 40:28Bonus
- 41:03Additional Resources
- 45:14Coordination
- 49:18Meaningful Operations
- 50:26Archiving
- 51:44Splitting
- 52:41Deleting
related talks
48mEvent Sourcing - what could possibly go wrong?
49mEvents... And what next?
01h 07mAn Architecture with CQRS and Event Sourcing
39mEvent Sourcing with Elixir
54mEvent Sourcing
52mEvent-driven vs event sourcing: what's really the difference?
43mEvent Sourcing dealing with an eventful past
52mEvent Sourced Transformation in Financial Systems