From my understanding regarding consistency :
Standby node on multi AZ : ACID consistency
read replica : BASE consistency (eventual consistency) due to aynsc replication
Am I right ?
thanks
2 Answers
Ehh, not really. We generally refer to BASE and ACID as broad characteristics of entire products rather than individual components in the products. S3 follows a BASE consistency model while RDS MySQL is ACID compliant.
Relational Databases like RDS are ACID compliant–even if their read replicas are async. Replicas should really just be thought of as extensions of the main database–all part of the same product.
In fact, to use Read Replicas with RDS MySQL, you must use the InnoDB storage engine, which specifically enforces transactional ACID properties.
thanks scott
Async replication is the base of eventual consistency ? the origin of this eventual consistency( for S3,….)
How RDS can guarantee ACID compliant if we are not sure to retrieve the same information everywhere at the same time ?
Not sure what you mean by async replication is the base of eventual consistency. Replicas are not a violation of ACID…they are replicas and we connect to them as separate endpoints. The big difference between ACID and BASE is how they can be updated. ACID compliant databases can only be updated on the master then changes get distributed to the read replicas. You can’t update a row on read replicas and have those changes flow back to the master. I can however, update S3 on any replica and it will become eventually consistent after a few seconds.