Roshi is a large-scale CRDT set implementation for managing timestamped events.
Roshi solves the problem of storing and querying large volumes of timestamped event data across distributed systems while maintaining consistency without requiring coordination. It uses conflict-free replicated data types to allow multiple nodes to accept writes independently and converge to the same state. The system is designed to handle scenarios where events arrive out of order or with conflicting timestamps, resolving these conflicts deterministically based on the CRDT semantics.
Roshi suits organizations operating at scale that need a distributed event store with eventual consistency guarantees. It is particularly valuable for use cases where strong consistency would create bottlenecks, such as tracking user activity, event logs, or time-series data across geographically distributed systems. The project is written in Go, making it suitable for deployment in modern infrastructure environments. Teams should adopt Roshi when they require a system that can absorb writes across multiple replicas without coordination overhead and can tolerate the read-after-write consistency model that eventual consistency provides.
The project shows sustained development activity with regular commits addressing bug fixes and incremental improvements. The codebase receives ongoing maintenance focused on stability and operational reliability rather than rapid feature expansion. Pull requests are reviewed and merged at a steady pace, indicating active stewardship. The project maintains clear documentation of its design decisions and operational considerations, reflecting a mature approach to distributed systems engineering. Issue tracking shows engagement with user-reported problems and clarifications on usage patterns.
