Coinbase CEO: Team to Reassess Architecture Trade-offs to Significantly Reduce Future Downtime

On May 9, Coinbase CEO Brian Armstrong posted on X platform stating, “Last night, Coinbase experienced a system outage, which should never have happened. The root cause of the failure was the simultaneous malfunction of multiple cooling units at an Amazon AWS data center, leading to excessive temperatures in the server room. We designed our services with redundancy to withstand the impact of any AWS availability zone (AZ) failure, and most systems operated normally under this mechanism last night, but not all systems were spared. The centralized exchange of the company could not withstand this failure. To achieve low latency and meet customer custody deployment needs, the exchange adopted a special architectural design. Although this theoretically allows the exchange to self-heal across availability zone failures, it increases transaction latency and undermines the advantages of customer custody deployment, which is counterproductive. After this incident, we will reassess these design trade-offs to strive to provide users with the most stable and high-quality trading environment. At the very least, we aim to significantly shorten system downtime during future availability zone switches. Thanks to AWS and the Coinbase technical team for working overnight to repair and mitigate the impact of the failure. We will compile a detailed technical incident review report and release it publicly once completed.”