Exam Professional Cloud Database Engineer topic 1 question 68 discussion

A. Google’s documentation (https://cloud.google.com/sql/faq) states MySQL read replicas use asynchronous replication. That eliminates B, C and D because they all say to use read replicas with synchronous replication. That leaves A. A states read replicas (plural) in a different region which suggests more than one read replica, each in a different zone. With the primary instance in region A (zone 1) and presumably a failover replica also in region A (zone 2), read replicas in region B (zone 1 and 2) would provide HA in the event of a region or zone failure.

C: Set up read replicas in different zones *** of the same region as the primary instance with *** synchronous replication, and set up read replicas in different regions with asynchronous replication.

no data loss means synchronous replication. Read replica in different zone will provide zonal level tolerance. Read replica in different region will provide regional tolerance.

synchronous replication will guarantee no data loss.

Cloud SQL for Postgres is using streaming replication which is asynchronous by default

Cloud SQL for Postgres is using streaming replication which is asynchronous by default

A. Because replication is asynchronous, when a regional outage occurs and a failover is attempted, some recent transactions that were committed to the primary may be lost (not replicated to the replica) - (https://cloud.google.com/sql/docs/postgres/replication/cross-region-replicas)

How is my data replicated? MySQL instances: MySQL instances provide a high availability configuration and MySQL read replicas. MySQL read replicas use asynchronous replication. PostgreSQL instances provide a high availability configuration and read replicas. SQL Server instances provide a high availability configuration and read replicas. https://cloud.google.com/sql/faq

Cloud SQL for MySQL doesn't directly support synchronous replication for read replicas, but third-party tools can be used for this purpose. Thus, C provides a better strategy for meeting the objectives.

Set up read replicas in different zones of the same region as the primary instance with asynchronous replication, and set up read replicas in different regions with synchronous replication. Synchronous is for data consistency and Asynchronous is for speed. So option D covers all aspects discussed.

The answer is A. Please see this link - https://cloud.google.com/sql/faq When it comes to read replica, there is no synchronous replication. How is my data replicated? MySQL instances: MySQL instances provide a high availability configuration and MySQL read replicas. MySQL read replicas use asynchronous replication.

If they meant for C to be a standby instance with synchronous replication and then a read replica in other regions with asynchronous replication then it would make sense. However, since it says read replicas only and read replicas do not get async replicated, then it should be A. Hard to say though because A isn't going to protect from data loss either. No really good answer here and badly worded question.

Vote C

I'm hung up on the phrase "without data loss". There is always a chance for some loss of data with asynchronous replication. However, Cloud SQL doesn't support synchronous replication on read replicas (as others have stated). There are NO VALID ANSWERS provided, and the authors of the question clearly have no idea what they're talking about.

Cloud SQL read replicas do not use synchronous replication. Instead, they use asynchronous replication. This means that there is a delay between when a change is made to the primary instance and when it is replicated to the read replica. The delay is typically a few seconds, but it can be longer depending on the network latency between the two instances. These leaves A at the end!

"without data loss" = "sync replication"

Non doubt