Ripple is a high-performance, distributed, in-memory event streaming system written in Go. Designed for scalability and fault tolerance, Ripple provides low-latency data streaming with support for replication, replay, leader election, and high-throughput consumer access.

• Distributed Architecture: Built for horizontal scalability with replication and leader election.
• In-Memory Streaming: Ultra-low latency event streaming, optimized for speed.
• Non-Blocking IO: Uses native OS-level IO multiplexing (epoll on Linux, kqueue on BSD/Unix).
• Fault Tolerant: Automatically detects node failures and re-elects leaders.
• Replay & Persistence: Events can be replayed for analytics or recovery.
• Replication: Asynchronous dumping to read replicas.
• Acknowledgment Levels: Supports various consumer acknowledgment modes.
• Load Balancing: Intelligent routing for reading and writing across nodes.
• gRPC API: High-performance RPC for communication.
• ZooKeeper Coordination: Leader election, metadata management, and cluster coordination.

The core server that handles incoming events, maintains in-memory queues, manages replication, and interfaces with ZooKeeper.

A Go client SDK for producing and consuming events from Ripple clusters.

Performance benchmarking tools to evaluate Ripple’s throughput and latency under different scenarios.

checkout benchmark for more experiments.