Coinbase CEO: Will reassess architectural trade-offs to significantly reduce future downtime duration

Coinbase CEO Brian Armstrong posted on the X platform that there was a service outage at Coinbase yesterday, which is absolutely unacceptable. The root cause was multiple cooling units at the AWS data center failing simultaneously, leading to overheating in one server room. Most of Coinbase's systems are designed with redundancy to handle the failure of a single availability zone, and they did operate normally last night.

However, centralized exchanges were not spared, as their architecture is specially optimized for low latency and customer co-location, making it difficult to simultaneously ensure fault tolerance at the availability zone level. While enabling the exchange to withstand availability zone failures would introduce latency issues and disrupt customer co-location, the team will reassess these trade-offs to at least ensure that when a switch to a different availability zone is needed, the duration of the outage can be significantly reduced. He expressed gratitude to the AWS and Coinbase teams who worked through the night to repair the issue and promised to release a detailed technical explanation later.