Which solution will meet these requirements MOST cost-effectively?
Use AWS CloudFormation templates. Deploy a stack with a DB cluster for each development group. Delete the stack at the end of each development cycle.
Use the Aurora cloning feature. Deploy a single development and test Aurora DB instance. Create clone instances for the development groups. Delete the clones at the end of each development cycle.
Use Aurora Replicas. From the primary writer instance, create read replicas for each development group. Promote each read replica to a standalone DB cluster Delete the standalone DB cluster at the end of each development cycle.
Use Aurora Serverless. Restore a current Aurora snapshot to an Aurora Serverless cluster for each development group. Select the option to pause the compute capacity on the cluster after a specified amount of time with no activity. Delete the Aurora Serverless cluster at the end of each development cycle.
Explanations:
Using AWS CloudFormation to deploy and delete DB clusters would automate the process, but it does not offer the cost savings of scaling down the database resources when not in use. Aurora DB clusters incur costs even when idle, making this solution less cost-effective.
Aurora cloning creates copies of the DB instance, but the clones are still full instances. This means the solution incurs costs for each cloned instance, which may not be cost-effective for a development environment that only needs a DB for 8 hours per day.
Aurora Replicas are read-only copies of the primary instance. Creating standalone DB clusters from read replicas incurs additional costs, and deleting these clusters after use is not as cost-effective as using a serverless solution.
Aurora Serverless can automatically scale compute capacity based on demand and pause during inactivity, making it the most cost-effective option for environments with infrequent use. Pausing the compute capacity reduces costs when not in use, and deleting the cluster at the end of each development cycle further reduces expenses.
I compute that the answer is:
Use Aurora Serverless. Restore a current Aurora snapshot to an Aurora Serverless cluster for each development group. Select the option to pause the compute capacity on the cluster after a specified amount of time with no activity. Delete the Aurora Serverless cluster at the end of each development cycle.