Welcome to FloRuntime
The Universal Distributed Runtime — Streams, KV, Queues, Time-Series, Actions, Workflows, and Stream Processing in a single binary. Thread-per-shard. Zero dependencies.
Flo is a single binary that unifies the data and orchestration primitives most systems stitch together from a half-dozen separate services. Instead of running Kafka for streams, Redis for KV, a queue broker, a time-series database, and a workflow engine — each with its own cluster, client, and operational surface — you run one process that speaks to all of them over a single wire protocol.
Everything in Flo is backed by the same Raft-replicated Unified Append Log. Writes are replicated and committed before they're acknowledged; reads are served from local projections with no cross-node round-trip. The runtime is thread-per-shard: each partition is owned by exactly one core, which gives predictable tail latency and lock-free hot paths.
Primitives
The data layer is four primitives, each tuned to a different access pattern but sharing one log, one consistency model, and one set of clients.
Orchestration
On top of the data layer, Flo runs durable execution and pipelines — no separate cluster needed.