AWS S3 was having issues with very high latency on the day of the outage. Additionally, AdvicePay was using a default HTTP client without a timeout to request firms' logo images from S3. The combination of those two issues led to a denial of service due to starvation of connections on our servers. AWS has since resolved things on their end. To protect against this issue in the future, AdvicePay has implemented timeouts for all HTTP clients that we use for requesting any resource.