Professional-Cloud-Database-Engineer Exam Dumps : Google Cloud Certified - Professional Cloud Database Engineer

In case of high CPU utilization like, mentioned in question, refer: "Check the CPU utilization of the instance. If the CPU utilization of the instance is above the recommended level, you should manually add more nodes, or set up auto scaling." Indexes and schema are reviewed post identifying query with slow performance. Refer :

"sub-millisecond latency" always involves Memorystore. Furthermore, as we are talking about a relational DB (Cloud SQL), BigTable is not a solution to be considered.

Recovery Process: Once Zone-B becomes available again, Cloud SQL will initiate the recovery process for the impacted read replica. The recovery process involves the following steps: 1. Synchronization: Cloud SQL will compare the data in the recovered read replica with the primary instance in Zone-A. If there is any data divergence due to the unavailability period, Cloud SQL will synchronize the read replica with the primary instance to ensure data consistency. 2. Catch-up Replication: The recovered read replica will start catching up on the changes that occurred on the primary instance during its unavailability. It will apply the necessary updates from the primary instance's binary logs (binlogs) to bring the replica up to date. 3. Resuming Read Traffic: Once the synchronization and catch-up replication processes are complete, the read replica in Zone-B will resume its normal operation. It will be able to serve read traffic and stay updated with subsequent changes from the primary instance.