The current implementation is a pull everything every 5 minutes. This would be useful to have it changed after the first pull to then query from the time that the last request was started. We may also need to have the exponential timeout backoff on failure as well.