Why Choose a Database Designed for K8s?
summary
Movatic migrated to EKS and attempted to run PostgreSQL in-cluster, then encountered two concrete failures: frequent pod restarts under write contention due to PostgreSQL being single-writer, and replication lag that made secondaries stale and caused failovers to lose recent rentals. They evaluated Kubernetes-native distributed SQL/NewSQL solutions (including CockroachDB), required a GitOps full-ops stack and an operator, performed extensive chaos testing (network latency, partitions, pod kills, cross-cluster faults), and selected a Kubernetes-native distributed SQL that preserves ACID under those failure scenarios.
AI-written summary · about
related talks
36mKube-native Postgres [I]
37mData On Kubernetes, Deploying And Running PostgreSQL And Patterns For Databases In a Kubernetes Cluster.
17mDebate: PostgreSQL on K8S
26mKubernetes from the Database Out
28mLightning Talk: Cloud Native PostgreSQL - Running PostgreSQL on Kubernetes
26mChallenges of Managing Stateful Applications in Kubernetes
26mBeyond the DBaaS Trap: Achieving Data Sovereignty With Kubernetes and CloudnativePG
25mPostgres on Kubernetes for the Reluctant DBA