Open-source Azure Files alternatives

SeaweedFS is a distributed storage system built to store very large numbers of files and serve them quickly. A single weed binary can provide an S3-compatible object store, a POSIX file system and S3 Tables for lakehouse use, all over the same data. The project is written in Go and released under the Apache-2.0 license.

Its design aims for constant-time disk access per file, and capacity grows by starting another volume server. Cloud storage can be cached or tiered transparently, and replication and erasure coding are supported. The same process can also run the master, volume servers, a filer, WebDAV, an Iceberg REST catalog and an Admin UI. You can start with one command, Docker, Docker Compose or Helm on Kubernetes, or build from source. The README includes comparisons with HDFS, Ceph, GlusterFS, MooseFS and MinIO.

Ceph is a distributed storage platform that offers object, block and file storage from one software-defined system. It is designed to scale across many servers and provide highly available storage, and it is written mainly in C++.

Repository topics show the range of protocols and features it covers: S3-compatible object storage, block storage, a POSIX distributed file system with FUSE access, protocols such as NFS, SMB, iSCSI and NVMe over Fabrics, and replication and erasure coding for data protection. The topics also mention Kubernetes and HDFS.

Most of Ceph is dual-licensed under the LGPL version 2.1 or 3.0, with some code in the public domain or under BSD-style licenses, which is why the repository lists the license as 'Other'. The documentation uses CC-BY-SA-3.0, and contributions need a Signed-off-by line without copyright assignment. Ceph is deployed on your own hardware or cloud infrastructure.

JuiceFS is a POSIX file system designed for cloud-native environments and released under the Apache License 2.0. Data written through it is persisted in object storage such as Amazon S3, while the matching metadata is kept in database engines such as Redis, MySQL or TiKV, chosen according to the scenario. It is written in Go.

Its aim is to let large amounts of cloud storage be used by big data, machine learning and AI platforms as efficiently as local storage, without code changes. Highlights include full POSIX compatibility, a Hadoop Java SDK compatible with Hadoop 2.x and 3.x, an S3 gateway, and a Kubernetes CSI driver for use in clusters.

The file system can be shared and read or written by thousands of clients, with strong consistency so that confirmed changes are immediately visible on all mounts. Data encryption in transit and at rest is supported, along with global file locks, both BSD flock and POSIX locks. A quick start guide is in the documentation.