Which solution meets these requirements?
Create and use a custom endpoint for the workload.
Create a three-node cluster clone and use the reader endpoint.
Use any of the instance endpoints for the selected three nodes.
Use the reader endpoint to automatically distribute the read-only workload.
Explanations:
Creating a custom endpoint allows for targeted read distribution among specific replicas. By directing queries to this custom endpoint, you can ensure that only the desired three replicas are used for the reporting queries, effectively distributing the workload according to the department’s needs.
A cluster clone is a separate instance of the original cluster and would not share the same replicas or resource specifications. This option would not automatically distribute queries across the existing replicas as required.
Using instance endpoints would target individual replicas directly, which does not provide automatic load distribution. This method is not suitable for ensuring that the reporting workload is evenly distributed across the selected three replicas.
The reader endpoint automatically distributes read requests across all available replicas, not just the selected three. This does not meet the requirement of directing the workload to specific replicas with different compute and memory specifications.
If I’m not mistaken, the answer is:
Create and use a custom endpoint for the workload.