The web app itself is completely isolated from the Lambda function doing the work, so that part can have downtime while the service keeps checking and sending a

The web app itself is completely isolated from the Lambda function doing the work, so that part can have downtime while the service keeps checking and sending alerts. On top of that, there's a bit of replication across regions (and soon across cloud vendors), so if an entire AWS region goes down (happened twice last month), the checks keep on checking.