How was it possible to double the RPS?

Doubling the RPS (requests per second) was achieved through a comprehensive approach:

• Optimization of request processing algorithms, which reduced response time.
• Implementation of caching at the application and database levels, reducing the number of heavy queries.
• Transition to asynchronous processing of certain tasks, which alleviated the main threads.
• JVM tuning: increasing heap size, optimizing garbage collection to reduce pauses.
• Horizontal scaling of the service using a load balancer.

For example, optimizing SQL queries and adding indexes allowed reducing query execution time from several seconds to tens of milliseconds, which directly impacted throughput.