I partially disagree. At a number of places I've implemented a, "Randomly, rarely, this request will be logged in detail into a detailed trace." The volume of

I partially disagree. At a number of places I've implemented a, "Randomly, rarely, this request will be logged in detail into a detailed trace." The volume of such traces can easily be kept down by limiting the frequency. But when we do this proactively, when a production problem happens, we can just look for a relevant trace. Alternately, at the end of the traced request, we can do something special. Like email the trace off if the request takes too long. This proved to be incredibly useful. But these days, you get this sort of capability easily with tools like Data Dog. And save yourself a lot of headaches in setting it up. So it is useful if your employer doesn't have that kind of monitoring pipeline. But if you don't have that, the manual version is a good thing to know.