Our command execution currently spans the time between ExecuteReader and the reader disposal; in other words, we report time-to-last-read. We also emit a received-first-response event in the span, which provides time-to-first-read information.
While this seems like the right thing to do by default, we may want to allow users to tweak this. For example, the received-first-response event may be unneeded information that can even cost money to store (see #4243). Also, some users may want the span to correspond to time-to-first-read instead. So we could have three modes:
- Span corresponds to time-to-last-read, and has the received-first-response event (the default, as today).
- Span corresponds to time-to-last-read, no received-first-response event.
- Span corresponds to time-to-first-read.
Our command execution currently spans the time between ExecuteReader and the reader disposal; in other words, we report time-to-last-read. We also emit a received-first-response event in the span, which provides time-to-first-read information.
While this seems like the right thing to do by default, we may want to allow users to tweak this. For example, the received-first-response event may be unneeded information that can even cost money to store (see #4243). Also, some users may want the span to correspond to time-to-first-read instead. So we could have three modes: