If you notice a sudden latency spike in a microservice, what systematic steps do you take to diagnose and resolve the bottleneck?
