The reporting workload is read-heavy, and the database shows high Read IOPS even outside the report window, suggesting sustained read pressure from other workloads or inefficient read patterns. The requirement is to improve both performance and availability of the RDS for MySQL instance. An RDS read replica is designed specifically to offload read traffic from the primary database instance and to provide additional capacity for read-heavy use cases such as reporting, analytics queries, and dashboards.

By deploying one or more read replicas, the company can direct the reporting job to a replica (Option B). This reduces contention on the primary instance, lowers read I/O demand on the writer, and can improve overall query latency and throughput. In addition, read replicas can contribute to availability objectives: if the primary instance has issues, replicas can be promoted (manually or as part of certain DR patterns) to become a new standalone database, reducing recovery time for read availability and providing a practical resilience option.

Option A (ElastiCache) can help for highly cacheable and repetitive queries, but it requires application/query redesign and cache invalidation strategy, and it does not inherently improve database availability. Option C is not valid because CloudFront is a CDN for HTTP content and is not an appropriate layer for database queries. Option D (vertical scaling) can improve performance, but it does not offload reads and often involves higher cost; it also does not provide the same availability and read scaling benefits as replicas.

Therefore, adding an RDS read replica and pointing the reporting workload to the reader endpoint best meets the performance and availability requirements.