rustfs-1-0-0-alpha-93

name	rustfs-1-0-0-alpha-93
description	High-performance distributed object storage system with S3-compatible API, OpenStack Swift support, and comprehensive observability features built in Rust. Use when deploying S3-compatible storage backends, configuring distributed clusters, integrating with Kubernetes via Helm, setting up TLS/mTLS, or building data lake solutions requiring high-throughput storage.
license	MIT
author	Tangled <[e-mail]>
version	0.1.0
tags	["object-storage","s3","swift","distributed-systems","observability","data-lake","rust"]
category	storage
external_references	["https://rustfs.com/","https://github.com/rustfs/rustfs"]

RustFS 1.0.0-alpha.93

Overview

RustFS is a high-performance, distributed object storage system built in Rust. It combines the simplicity of MinIO with the memory safety and raw performance of Rust, delivering full S3 compatibility under the permissive Apache 2.0 license. Unlike competitors using AGPL licensing, RustFS avoids legal restrictions while providing enterprise-grade features for data lakes, AI/ML workloads, and big data pipelines.

Key characteristics:

• Rust-based: Memory safety by design with C/C++ level performance
• S3 Compatible: Full AWS S3 API support including S3 Select
• OpenStack Swift: Native Swift protocol with Keystone authentication
• Distributed Architecture: Decentralized peer-to-peer design, no single point of failure
• Apache 2.0 License: Permissive licensing for unrestricted commercial use
• No Telemetry: Full data sovereignty, GDPR/CCPA/APPI compliant
• Observability: Built-in OpenTelemetry integration with Prometheus, Grafana, Jaeger, Tempo

When to Use

• Deploying S3-compatible object storage on-premises or in private cloud
• Replacing MinIO with a permissively licensed alternative (drop-in binary replacement)
• Building data lake infrastructure for Spark, Presto/Trino, Iceberg, Hudi, Delta Lake
• Storing AI/ML training datasets and model artifacts at scale
• Implementing cold archiving or long-term data retention with WORM support
• Setting up cross-region active-active replication for disaster recovery
• Integrating object storage with OpenStack via Swift API and Keystone
• Deploying lightweight edge storage (binary under 100 MB)
• Replacing HDFS with a cloud-native, multi-protocol alternative
• Configuring observability pipelines with Prometheus metrics and distributed tracing

Core Concepts

Object: The fundamental unit of storage — files, byte streams, or any unstructured data. Maximum size is 5 TiB via multipart upload.

Bucket: A logical container for objects. Data is isolated between buckets. Similar to a top-level directory from the client perspective.

Drive: The physical disk storing data, passed as a parameter at startup. All object data resides on these drives. JBOD mode is recommended (no hardware RAID).

Set (Erasure Set / Stripe): A group of drives distributed across different nodes. An object is stored within a single set. The cluster automatically divides into sets based on scale. One object = one set. One cluster = multiple sets.