How would you approach a situation where a database service is experiencing latency spikes at Datadog?
