At around 2:00 AM on July 31 our Cloud provider, Amazon AWS, had trouble with three of the services that we use (SQS, SNS, and SES) simultaneously. The outage lasted until about 3:40 AM. This impacted about 50 transactions. We have been able to recover all but 3 of those transactions, and have contacted those individuals to help them complete their transactions.
Our design had relied on using each of these services as backup for the others. So if the Queue service was down, we used the Notification or Email Service and vice versa.
We have now deployed an additional backup in case this ever happens again by essentially saving transactions to file storage in the event that all three services are down again.