100% database utilization every second day continuing for several days

Hi,

I have moved a local teamcity server installation into azure cloud on a virtual machine and using an azure SQL DB. 

I have noticed that every now and then the database is fully utilized for 2 days, it seems to happen 2 days with normal usage and approx 2 days fully utilized (using a standard S0 10DTU DB). When the DB is fully utilized teamcity becomes unresponsive, I have let teamcity finish whatever it is doing to see if it was optimization or clean up, but it continues happening. If I restart the teamcity service, everything works as expected afterwards, so it doesn't seem to be the DB that is acting up as it would be expected that after the restart it would still be a problem then. 

The below screenshot shows the periodic behavior.

I have looked at the logs: cleanup, diagnostics activities, maintenance and server log and cannot see anything happening when this starts occurring. 

Drilling down in the sql log of teamcity doesn't give me any insights other than starting to notice queue takes longer to execute. 

This does not seem to be related to IO either, looking at below graph it can be see that IO doesn't seem to peek at that specific time when the problem occurs.

Looking into the specific queues that are triggered often, it seems like they triggers less often due to them taking longer time but they increase in total time running. Example:

 (13:45:34, global_settings) 1 - [1ms] SQL SELECT: select top(1) id, task_type, description, stage, node_id, created, updated from config_persisting_tasks where task_type = ? and stage in (?,?,?) order by id | PARAMETERS: "global_settings", 1, 2, 6

The execution count increases after the DB is back to normal and totoal duration decreases.

I currently doesn't have the teamcity sql log file from when this started happening, as they have all been overwritten.

Does anyone have any clue to what might be going on or any suggestions where to go from here ?