The Most Difficult Observability Tool?

The decompression of that compressed like playback set is not compute intensive. It's scheduled by the browser before we actually receive it. The pipeline for which data kind of flows through our system is a little bit unique compared to, for example, reporting errors and enhancing them with stack traces or things like that. We have a whole sort of like exponential back off type mechanism on the client so that if requests fail, we make sure that data gets across. After that, once stuff goes into Postgres and a session,. let's say, is over, then we throw it into long term storage, something like S3.