SeaweedFS

lastmod 2026-03-12created 2022-05-135 edits

SeaweedFS is a fast distributed storage system for blobs, objects, files, and data lake, for billions of files! Blob store has O(1) disk seek, cloud tiering. Filer supports Cloud Drive, cross-DC active-active replication, Kubernetes, POSIX FUSE mount, S3 API, S3 Gateway, Hadoop, WebDAV, encryption, Erasure Coding.

A simple and highly scalable distributed file system
The goal is to store billions of files and serve them fast
Implemented with ideas from Facebook's Haystack design and f4: Facebook's Warm BLOB Storage System
The Blob store has O(1) disk seek and cloud tiering
The Filer supports directories and POSIX attributes, and includes Cloud Drive, cross-DC active-active replication, Kubernetes, POSIX FUSE mount, S3 API, S3 Gateway, Hadoop, WebDAV, encryption, Erasure Coding, and more

Demo

docker run -it --rm -p 7333:7333 -p 8333:8333 chrislusf/seaweedfs server -s3 -webdav

s3 is at localhost:8333
webdav is at localhost:7333

Features

Offers various replication options, such as replication level and rack- and data-center-awareness.
Automatic master server failover means no single point of failure.
Supports automatic Gzip compression based on file MIME type.
Automatic compaction to reclaim disk space after data deletion or update.
Total storage space can be increased by adding servers.
No data rebalancing when adding/removing servers; only triggered by an administrator command.
Supports image resizing, ETag, Accept-Range, Last-Modified, etc.
Supports tuning in-memory/leveldb/readonly modes to balance memory and performance.
Provides customizable multiple storage tiers and transparent cloud integration.
Rack-aware 10.4 erasure coding for warm storage reduces storage cost and improves availability.

Filer features

The Filer server serves "normal" directories and files over HTTP.
File TTL automatically expires file metadata and the actual file data.
Files can be read and written directly through a filer mounted via FUSE.
Filer store replication enables HA for the metadata store.
An Amazon S3-compatible API allows access to files using S3 tools.
A Hadoop-compatible file system allows access to files from Hadoop/Spark/Flink, etc.
AES256-GCM encrypted storage stores data securely.
Supports storing large files of tens of TB.
Cloud Drive mounts cloud storage to the local cluster for fast reads and writes.

Kubernetes

Supports a Kubernetes CSI driver and the SeaweedFS Operator.

Seaweed object store usage examples

Explains the default ports and how to run master nodes and volume nodes.
Provides examples of how to write, read, and delete files.
Explains rack-aware and data-center-aware replication, and how to assign file keys to a specific data center.

Architecture

SeaweedFS manages data volumes instead of splitting files into chunks.
The master server manages only metadata about volumes, while the actual file metadata is managed by the volume servers.
Provides a detailed explanation of the file write and read processes.

Comparison with other file systems

Explains SeaweedFS's advantages compared with HDFS, GlusterFS, Ceph, MooseFS, MinIO, etc.
SeaweedFS is optimized for small files and provides fast file access with O(1) disk read operations.