Schema proliferation builds slowly and gets expensive fast.
One schema per event type seems reasonable - until you're: • Managing 10+ tables • Writing union queries across all of them • Propagating a single field rename everywhere
The Alternative❓ Discriminator-based schema consolidation.
It reduces schema count, simplifies downstream consumption, and makes schema evolution manageable. New variants become additive changes instead of breaking existing consumers.
🔗 Check out the #InfoQ article for a practical look at the pattern and its trade-offs ⇨ https://bit.ly/4uC023n
Comments (0)