Two days ago, with no code changes or changes to the DB, I am not getting a lot (every 5 minutes or so) of errors with The wait operation timed out
error with t
We resolved this issue, along with other types of random timeouts on SQL Azure by switching to "contained users". Using server-level logins on SQL Azure can cause issues:
This is not very efficient as in SQL DB master and user can sit on two different SQL servers potentially in two different machines. Also when a server has multiple user databases then master will be the bottleneck in the login process, and under load this may result in high response time for logins. If Microsoft is updating the software on the machine / server then master will be unavailable for a few seconds and all the logins to the user database can fail too at this time (http://www.sqlindepth.com/contained-users-in-sql-azure-db-v12/)
As in your case, I had my doubts because my database was not under heavy load, but switching to contained users made a tremendous difference anyway.
The SQL to create these users is as follows (run this on the database itself, not on the master database as you would for creating server-level logins):
Create user ContainedUser with password = 'Password'
ALTER AUTHORIZATION ON SCHEMA::[db_owner] TO [ContainedUser]
ALTER ROLE [db_owner] ADD MEMBER [ContainedUser]