What is the mechanism of “Avg Replica Lag” metric in Cloudwatch?

Question

For Amazon RDS, there is a metric called "Avg Replica Lag" in Cloudwatch. The meaning is quite clear. It stands for average replication delay between the master to slaves.

However, I'm not sure the mechanism of this metric, i.e. how does Amazon RDS detect such a lag? Taking MySQL database as an example, does Amazon RDS have some specific methods to detect the delay? Or it simply uses the results of "second behind master" reported by MySQL?

Answer 1

Cloudwatch's Replica Lag is the average of the results returned by "Seconds Behind Master" when you perform a 'SHOW SLAVE STATUS' query on all of your slaves.