A load balancer server failed causing a portion of requests to fail or timeout. Requests were automatically routed to alternate datacenter. We are looking at multi-region load balancers to avoid this issue in the future.