The main purpose of employing read replicas in a database architecture is to increase the application's read throughput and to provide data locality for read queries, which can result in lower latency. Read replicas allow you to scale out beyond the capacity constraints of a single database instance for read-heavy database workloads. By doing so, you can reduce the load on the primary instance thereby improving the overall performance of your database. It's important to note that read replicas are not used for increasing write capacity because they are read-only and do not accept write operations.