Which method can improve relational database read performance?

Adding a read replica to a relational database can significantly improve read performance. A read replica is a copy of the database that can handle read requests, allowing the primary database to focus on write operations. By distributing read traffic across multiple replicas, the overall workload is lowered for the primary database, reducing the likelihood of bottlenecks and improving response times for read queries. This is particularly beneficial in environments with high read demand, as it enhances scalability and reliability.

In contrast, vertical scaling involves adding more resources (such as CPU and RAM) to a single database instance, which may provide some improvements but has limits regarding how much performance can be enhanced. Auto-sizing pertains to the dynamic adjustment of resources based on usage patterns, which helps optimize performance but may not specifically target read operations. Scoping horizontally generally refers to distributing load across multiple services or components rather than specifically improving read performance within a single relational database instance.