bucket4j

Uses integer calculations exclusively to avoid floating-point rounding errors, ensuring accurate token consumption and refill rates as highlighted in the README.

Default lock-free implementation provides excellent scalability for multi-threaded environments, reducing contention and improving performance in high-concurrency scenarios.

API optimized with primitive types minimizes boxing and garbage collection, making it suitable for performance-critical applications as stated in the features.

Supports numerous backends like Redis, JDBC, and MongoDB for clustered rate limiting, with async APIs for non-blocking operations in distributed setups.