Chronicle Ring
Chronicle Ring provides a high-performance, multi-writer, multi-reader ring buffer implementation designed to reduce tail latencies and eliminate data copying. It offers a specialised "RingZero" version optimised for single-writer, single-reader use cases, delivering sub-microsecond write-to-read latencies for the most demanding financial and high-throughput applications.
Talk to Our TeamKey Features
Ultra-Low "Zero-Copy" Latency
By directly accessing the underlying ring bytes through a BytesCallback interface, Chronicle Ring eliminates the overhead of data copying, ensuring the lowest possible write-to-read latencies.
Flexible Implementation Models
Choose between BlockingRingZero for guaranteed, ordered message delivery, or ConflatingRingZero for market data scenarios where the reader only needs the latest available update.
Cross-Process & In-Process Communication
Supports Named instances backed by memory-mapped files for persistence and inter-process communication on the same host, or Anonymous instances for high-speed messaging between threads in a single process.
Optimised for Single-Writer/Single-Reader (RingZero)
The specialised RingZero implementation is tailored for the common 1:1 messaging pattern, featuring a fixed number of slots (power of 2) and fixed maximum message sizes to maximise throughput.
Small, Intuitive API Surface
The interface is kept intentionally small, consisting of just two primary calls (writeHandle and readHandle), and is designed to be used with standard Java try-with-resources for clean lifecycle management.
Proven Performance Baseline
Engineered for predictability, Chronicle Ring can achieve latencies as low as ~150ns at the 50th percentile and remains under 800ns even at the 99.999th percentile on standard enterprise hardware.