We encountered similar problems in the past using SQL Azure; often queries work with tables with less than 10 rows and even standard queries of the .Net membership provider, all the time interrupted with timeouts. This usually happens when we have little activity on our service; mostly at night.
In frequently used areas where you can safely repeat the SQL timeout request (usually read operations), we have added a timeout exception to our own error detection strategy, taken from the Transient Error Handler ; however, as you said, this is not suitable in most cases.
The best explanation we have received from Azure support so far is that SQL Azure is indeed a shared instance of SQL Server that is used by multiple clients; if one user performs an intensive operation, this may affect other users. But; Consider this unacceptable, we are still in contact with SQL Azure support to find out why throttling does not stop this activity from affecting us.
The best thing:
- Contact Azure SQL Support through the forums or directly (if you have a support package)
- If it is possible; try setting up a new instance of SQL Azure and migrate the database through
- While this problem occurs intermittently on one instance of SQL Azure; we have never experienced this in our other two cases.
As a side note; we are still waiting for Azure support to get back to us regarding why we were still receiving exceptions while waiting.