We know what it’s like when the world’s on fire at 3 am and your tools are lying to you. Or when everything seems fine, but your most important customer is insisting there’s a huge problem.
Get results on the fly, zoom down to the level of each individual requesti, and break it all down by arbitrarily fine-grained attributes like
raw_query. Never download a raw log from S3 again.
Use the context and history from previous debugging sessions to isolate complex root causes. Go back and see how similar symptoms were tracked down and resolved in the past.