Monitoring SGA utilization

Is there anyway I could schedule a job or is there any scripts which I could scheduled into the scheduler to collect SGA utilization on an hourly basis?

I'm having some very weird behaviour on my production databases where the application disconnects almost daily at a certain time of a day. I cannot pin point which script was run during the time when disconnection occurs as it's different everytime.

I do notice that whenever application connection disconnected from the DB, I see an abnormally high latch: shared_pool waits from the AWR. Multiple tickets has been raised to MOS but yield no results at all. This is because there is no error detected in either the DB alert logs or even listener logs. Application logs only says connection to DB lost.