distributed-systems

etcd-io/etcd

The Raft-backed key-value store behind Kubernetes — etcd is a distributed key-value store using Raft consensus, offering a gRPC API, automatic TLS, and benchmarked 10,000 writes/sec. It underpins Kubernetes and other critical systems, with rigorous robustness testing backing production reliability for teams needing consistent, highly-available coordination data.

spacedriveapp/spacedrive

Spacedrive aims to make self-owned storage practical across devices — Spacedrive is building a virtual distributed filesystem so your files stay yours across devices instead of living on someone else's server. A 2.0 beta planned for November 2026 adds portable indexes, private remote access, mounts, history, and extensions. Nightly CLI/daemon builds are already available for testing, aimed at people wanting self-hosted, cross-platform file management.

seaweedfs/seaweedfs

One Go binary unifying S3, POSIX, and Iceberg lakehouse storage — SeaweedFS combines an S3 object store, a POSIX file system, and a lakehouse with S3 Tables in a single 'weed' binary over shared data. It targets billions of files with O(1) disk access, horizontal scaling via extra volume servers, and transparent tiering to cloud storage.

temporalio/temporal

Durable execution server for resilient, retrying workflows — Temporal runs application logic as Workflows that automatically survive failures and retries, forking from Uber's Cadence into a mature, independently developed platform. It suits teams building reliable distributed systems or microservice orchestration who want durable state without hand-rolled retry logic.

nats-io/nats-server

A lightweight messaging server built for cloud, edge, and IoT scale — NATS is a CNCF project offering a single Go server that handles pub/sub, request-reply, and streaming for distributed systems. With over 40 client language bindings and the ability to run anywhere from cloud clusters to a Raspberry Pi, it suits teams building microservices or connected-device architectures that need simple, secure messaging.

juicedata/juicefs

A POSIX Layer That Turns S3 and Redis Into Shared Storage — JuiceFS lets S3-style object storage act like a full POSIX filesystem, with metadata kept in Redis, MySQL, TiKV or similar engines. It connects cloud storage to big data, ML and AI pipelines without code changes, making it useful wherever Hadoop-style HDFS workloads need to move to the cloud.

encoredev/encore

Declare infrastructure in code, run it locally, deploy to AWS or GCP — Encore lets you declare databases, pub/sub topics, and object storage directly in your backend code via an SDK, then run the full stack locally with its CLI and auto-provision matching cloud resources on deploy. Built for developers and AI coding agents who want to skip manual infra setup while still being able to fall back to Terraform for production.

git-bug/git-bug

Bug tracking that lives entirely inside your git repo — git-bug stores issues directly in git objects, so tracking works offline and syncs through normal git push/pull without extra servers. It offers CLI, terminal UI, web UI and a GraphQL API, plus bridges to import/export data from other trackers—useful for teams wanting to avoid vendor lock-in.

paradigmxyz/reth

A Rust-built Ethereum execution client made for hacking on its internals — Reth is a production-ready Ethereum execution layer client written in Rust, built by Paradigm as a set of modular, reusable libraries rather than a monolith. It works with any Engine API-compatible consensus client, making it appealing to developers who want to customize or build on top of individual components.

hydro-project/infinity

Rust runtime lets 75k agents hibernate until their next event — Infinity is a Rust framework for massively concurrent agentic systems, fitting 75k agents in a Raspberry Pi's memory. Agents hibernate between turns as plain conversation data with no stack or open connection, then wake on the next message. The same code runs on a resident process or scales to zero on AWS Lambda.

salihcantekin/ruvio

In-memory key-value server in Rust that speaks RESP, so redis-cli and Redis clients connect. Locks, rate limits, job queues and a change feed are single commands; Raft-based failover. Clients for Rust, .NET, Go, Java, Python and Node.js.

anoneurx/lynx

LYNX Builds Its Own Search Index Instead of Renting Bing or Google's — LYNX is a Rust-based search engine that crawls, indexes, and ranks (BM25F) entirely on its own infrastructure rather than syndicating results from Bing or Google like most privacy-focused alternatives. v0.1.0 covers the full pipeline end-to-end, including an optional AI answer layer, and targets developers and privacy-conscious users wanting a self-hosted, inspectable search stack.