Each Google Cloud SQL instances has a single MySQL instance at any one time. The data is replicated to multiple locations when that single MySQL instance writes it to persistent storage -- this means that the instance can failover to a new location if there is a problem.
There isn't any hardware clustering in the sense used here.