Which solution will meet these requirements MOST cost-effectively?
Scale the existing production database in a maintenance window to provide enough power for the data scientists.
Change the setup from a Single-AZ to a Multi-AZ instance deployment with a larger secondary standby instance. Provide the data scientists access to the secondary instance.
Change the setup from a Single-AZ to a Multi-AZ instance deployment. Provide two additional read replicas for the data scientists.
Change the setup from a Single-AZ to a Multi-AZ cluster deployment with two readable standby instances. Provide read endpoints to the data scientists.
Explanations:
Scaling the existing database increases costs and impacts production performance, which may not meet the availability and isolation needs of the data scientists.
In a Multi-AZ deployment, the standby instance is for failover and is not readable, so this setup does not allow data scientists access to the standby instance for queries.
Adding two read replicas in a Multi-AZ deployment can provide read access for data scientists, but this option incurs unnecessary costs for multiple replicas, reducing cost-effectiveness.
A Multi-AZ cluster with two readable standby instances provides high availability and isolated read-only access for data scientists at a more cost-effective scale.
I weigh that the answer is:
Change the setup from a Single-AZ to a Multi-AZ cluster deployment with two readable standby instances. Provide read endpoints to the data scientists.