Today we have a default time out on query operations (ExecuteSql, ExecuteStreamingSql, Read, StreamingRead) of 60 seconds. We have instances of users authoring queries that take longer than this and being surprised by a 504 DEADLINE EXCEEDED. I propose:

1. improving the error we throw when the deadline is caused client, not server side, to point the user to look into timeout and retry configuration.
2. Consider adding an intermediate timeout setting for these methods. Today we have Default (60s) and long_running (1 hour). A setting that was something between, like 5 minutes, could alleviate a higher % of users finding this limit in the first place.