Day 8 of #DistributedDilemmas: Eventual consistency is consistently misunderstood
Eventual consistency is consistently misunderstood. CRDTs work better in whitepapers than in production. Real-world systems need a spectrum of consistency…
CRDTs work better in whitepapers than in production. Real-world systems need a spectrum of consistency models!
Latency vs. Accuracy Trade-offs:
-
PayPal’s regulatory compliance platform adopted CRDTs for geo-distributed transactions but faced challenges with exactly-once delivery and causality tracking when resolving payment conflicts. https://www.infoq.com/presentations/crdt-production/
-
aerospikedb highlights that eventual consistency risks data staleness and requires complex conflict resolution mechanisms (e.g., vector clocks) to handle concurrent updates. https://aerospike.com/glossary/eventual-consistency/
-
TiDB_Developer uses CRDTs for IoT data aggregation but combines them with strong consistency for critical operations like financial transactions. https://www.pingcap.com/article/understanding-crdts-and-their-role-in-distributed-systems/

Keep reading
🚀 Launching today: 28 Days of #DistributedDilemmas
Follow me on Twitter/X this February for one battle-tested insight a day on the hidden complexities behind outages, tech debt and architectural nightmares.
#DistributedDilemmas: all 24 days in one place
In February 2025 I posted one distributed systems dilemma a day on X. I made it to day 24. Here they all are, with the diagrams.
Cell-based architecture, explained: the public version of my Innovation Day talk
My Innovation Day deck was internal, so here's the public version: what a cell is, scaling up and across, blast radius, and what cells cost you.
Leave a comment