Monitoring Availability Groups can be tricky. The DMVs don’t update the log_send_rate and redo_rate all the time especially if nothing is happening so if you try to use those for calculations when monitoring you could false results worse yet pages in the middle of the night. In order to calculate the log_send_rate and redo_rate you need to capture the perfmon counters ‘Log Bytes Flushed/sec’, ‘Redone Bytes/sec’, and ‘Log Bytes Received/sec’ into temporary tables WAITFOR 1 second then capture them again. Below is query that captures this along side what you see in the DMV for when Microsoft might fix the issue.
Click through for a script as well as a Powershell cmdlet wrapper for running against a set of hosts in your Central Management Server.