The solution wasn't to add more instances or tweak timeout values. We needed to fundamentally change how services communicated. We introduced an event-driven architecture using Azure Service Bus and ...