Which solution will meet these requirements?
Recreate the cluster with a shared cluster volume. Add two instances to serve both read requests and write requests.
Add one read replica instance. Activate a shared cluster volume. Route all read queries to the read replica instance.
Add one read replica instance. Set the read preference to secondary preferred.
Add one read replica instance. Update the application to route all read queries to the read replica instance.
Explanations:
Recreating the cluster and adding instances will increase costs and complexity, but it doesn’t directly address the need for improving performance based on read-heavy workloads.
Shared cluster volume is not a feature of Amazon DocumentDB. This solution would not provide the intended performance boost.
Adding a read replica and setting the read preference to “secondary preferred” allows read traffic to be routed to the replica, thus improving performance without modifying the application.
Updating the application to route read queries manually adds overhead and is less optimal compared to automatically routing reads via read preference.
In my experience, the answer is:
Add one read replica instance. Set the read preference to secondary preferred.