onJune 15, 2026

RustFS Brings Native Support for S3 Tables: Upgrading Object Storage to AI-Native Storage

Security

3 min read

Nearly a year since its open-source debut, RustFS has achieved full compatibility with the S3 protocol, becoming the world’s fastest-growing distributed object storage system. Today, we are excited to announce that native support for Amazon S3 Tables is nearing completion. Released under our ongoing open-source commitment, this milestone officially elevates RustFS from traditional object storage to AI-native storage.

S3 Tables: The Storage Imperative for the AI Era

In 2006, AWS launched S3, pioneering a new era for managing unstructured data (video, images, text, etc.) and establishing the de facto standard for object storage.

In 2024, AWS introduced S3 Tables. By debuting a new “Table Buckets” architecture, AWS extended S3’s boundaries directly into structured data. For the first time, the management of both structured and unstructured data has been unified at the object storage layer.

This paradigm shift is driven by the rapid advancement of AI and the urgent demand for efficient, large-scale data management.

Data is the lifeblood of the AI era. Large Language Model (LLM) training demands massive volumes of both structured and unstructured data, while inference and Agent workflows continuously generate immense datasets. However, traditional storage architectures typically relegate structured data to standalone databases or data warehouses, while leaving unstructured data to object storage or file systems. In an AI era that demands extreme data mobility and efficiency, this fragmented architecture introduces critical bottlenecks:

Operational Complexity: Organizations must maintain multiple heterogeneous storage systems. As data from LLM training and inference scales to petabyte (PB) or exabyte (EB) levels, the operational overhead and resource expenditures of managing disjointed systems grow exponentially.
Data Processing Inefficiency: AI applications frequently context-switch between heterogeneous systems to exchange data. This cross-system data movement hogs valuable system I/O and network bandwidth, introducing massive latency that severely slows down end-to-end data pipelines.
Lack of AI-Native Design: Traditional storage architectures lack semantic awareness for AI-specific workloads (e.g., RAG pipelines and Agent memory storage). Because structured and unstructured data remain siloed, LLMs struggle to perform direct cross-modal joint retrieval and reasoning. Consequently, enterprises are forced to build complex middleware to bridge the gap between underlying storage and AI business logic.

S3 Tables break this deadlock. By embedding data lake management capabilities (such as the Apache Iceberg table format) directly into the object storage core, it dismantles the silos between heterogeneous data systems. Offering a standardized data management interface, it delivers true AI-native capabilities to the storage layer.

How RustFS Implements S3 Tables Support

With full S3 compatibility already under its belt, RustFS seamlessly integrates with any system supporting the S3 protocol. By extending these capabilities to S3 Tables, our implementation delivers distinct advantages:

Open and Extensible: RustFS is distributed under the business-friendly Apache 2.0 license. S3 Tables support will be baked directly into the core and shipped under the same open-source protocol, ensuring complete freedom and flexibility for our users.
Powered by Rust: Leveraging Rust’s memory safety and high-concurrency performance, RustFS handles the intensive metadata operations brought by S3 Tables (such as snapshot manifests and file compactions) with superior stability and efficiency—perfectly matching the rigorous, low-latency demands of AI training workloads.
Enterprise-Grade Security: RustFS features a comprehensive, built-in security suite, including full-lifecycle IAM management, security auditing, mTLS, and KMS. These mature capabilities will be fully injected into S3 Tables to ensure robust data governance.
Frictionless Deployment: RustFS supports versatile installation paths (source compilation, pre-built binaries, containerization) and cross-platform deployments (Linux, macOS, Windows, NixOS, etc.). Users can spin up a RustFS instance with a single command and jump straight into S3 Tables.
Unified Data Management: In AI scenarios like RAG, both raw unstructured data and the processed, structured data湖 (data lake) metadata can be unified within RustFS. Through standard table format interfaces, upstream AI engines can seamlessly execute joint analysis and retrieval, eliminating tedious data synchronization between separate object stores and analytical databases.

Supporting S3 Tables is a pivotal step in our mission to build the definitive storage infrastructure for AI data centers. By bringing the entire lifecycle of structured and unstructured data under a unified RustFS ecosystem, we drastically cut data management costs for LLM training, inference, and AI applications, while maximizing storage throughput and space utilization.

About RustFS

RustFS is a next-generation distributed object storage system written in Rust, purpose-built for AI data centers. Open-sourced under the Apache 2.0 license, RustFS has amassed over 28k GitHub Stars in just 10 months, trending 9 times on the global GitHub Trending list and 50 times on the GitHub Rust Trending list.

To date, RustFS boasts a global deployment base of over 1 million instances and serves multiple global enterprise customers. In April 2026, RustFS officially joined the NVIDIA Inception global accelerator program. Leveraging the NVIDIA DPU ecosystem, RustFS is actively building a next-generation storage operating system optimized for DPUs.