A single data center can't withstand quorum loss; Coinbase's lesson was costly enough.

Crypto World News reports that Coinbase released a report stating that on May 7, 2026, service was interrupted for about 8 hours due to a cooling system failure in the AWS us-east-1 region use1-az4, with full recovery taking approximately 12 hours. The trading matching engine lost quorum due to a single data center architecture and needed code adjustments and new node groups to resume operation. AWS-managed Kafka service control plane failure further expanded the impact. Coinbase will upgrade the cross-region hot standby architecture, migrate Kafka to a three-availability-zone deployment, and work with AWS to promote improvement measures.