Obstor

Obstor is an open-source object storage server for teams that want S3-compatible storage they can operate themselves, with SFTP as another way to move files in. It is most compelling when you can run a distributed deployment; its standalone mode is intended for development and evaluation.

Its S3 compatibility and multi-site replication make it a flexible self-hosted option, but versioning, object locking, and bucket replication depend on distributed deployment with erasure coding. That is an important architectural commitment, not a minor configuration detail.

Overview

Obstor works with S3 SDKs, CLI tools, and applications without code changes, so teams can keep existing S3-oriented workflows while running the storage server on their own infrastructure. Native SFTP uploads are stored as S3 objects, bringing file transfers into the same object store rather than creating a separate storage path.

In distributed deployments, objects are automatically sharded and replicated across connected nodes, with each object stored in at least two global locations. Continuous replication and failover orchestration support multi-site operation, while erasure coding with configurable parity helps protect against drive and node failures. These are meaningful strengths for teams building resilient storage, but they make deployment design central to getting the capabilities that matter.

Key features

Access, protection, and management

Obstor supports SSE-C and SSE-S3 server-side encryption, including KMS integration. Its authenticated encryption uses AES-256-GCM or ChaCha20-Poly1305, with HMAC-SHA-256 for key derivation. IAM users, groups, policies, and service accounts use S3 and AWS IAM policy syntax, which suits administrators who already work with those access-control conventions.

Versioning and lifecycle policies support object history, recovery, and retention requirements, while event notifications can be sent to Kafka, NATS, AMQP, Redis, PostgreSQL, Elasticsearch, or webhooks. The key caveat is topology: standalone servers are for development and evaluation, and versioning, object locking, and bucket replication require a distributed deployment with erasure coding. Teams that need these safeguards should plan for a distributed configuration from the start.

Deployment choices

Obstor can be installed as a single binary, through Docker, with a Kubernetes Helm chart, or from source. Official standalone instructions cover macOS, GNU/Linux, and 64-bit Windows. The Docker Compose example creates four distributed instances and supports up to 16 Obstor services, giving operators a defined route to a multi-instance setup. Those options offer flexibility, but the standalone instructions should not be mistaken for a production-ready substitute for distributed deployment.

Pricing

Open source — free. The plan is self-hosted and licensed under Apache 2.0, making it a fit for teams able to supply and operate their own infrastructure. There is no paid tier or hosted service described in this plan; the trade-off is that deployment and operations remain with the user. The standalone limitations also matter: teams needing versioning, object locking, or bucket replication must use distributed deployment with erasure coding.

Platforms

Obstor supports API, Linux, macOS, self-hosted, web, and Windows platforms. Official standalone instructions are available for macOS, GNU/Linux, and 64-bit Windows; deployment options also include Docker and a Kubernetes Helm chart. This range fits infrastructure teams deploying across common operating systems, containers, or Kubernetes, rather than readers looking for a conventional end-user app.

Who it's for

Obstor is best for operators and development teams that want an S3-compatible object store under their control, need SFTP access alongside S3 workflows, and can manage distributed infrastructure. The project describes machine learning, analytics, and application data as suitable workloads, and its replication, access-control, encryption, and notification features support those use cases.

It is a weaker fit for anyone who needs a simple standalone production store with versioning or bucket replication, or who does not want to run storage infrastructure. Those needs either require the distributed deployment or point toward a different operating model.

Pros and cons

• Pro: S3 SDKs, CLI tools, and applications can connect without code changes, easing adoption in existing workflows.
• Pro: Native SFTP stores uploads as S3 objects, so file transfer and object access converge on one store.
• Pro: Distributed sharding, replication across at least two global locations, and configurable erasure coding address multi-site placement and infrastructure failures.
• Pro: Apache 2.0 licensing and multiple deployment routes give self-hosting teams flexibility without a software fee.
• Con: Standalone mode is intended for development and evaluation, not as the route to the full production feature set.
• Con: Versioning, object locking, and bucket replication require distributed deployment with erasure coding, increasing operational commitment for teams that need them.
• Con: The free plan is self-hosted, so infrastructure and operations are the user's responsibility.

Alternatives

For a broader category comparison, browse Storage Replication Software.

• Synology Photos is a free alternative for photo-focused use when you have a Synology NAS; it supports mobile, desktop, web, and browser-extension platforms rather than serving as an S3-compatible object store.
• SeaweedFS is another self-hosted option across API, Linux, macOS, and Windows, with a free community-support plan and an available free trial; choose it for a different open-source storage project.
• LINBIT DRBD offers a free open-source community version for Linux and self-hosted environments, with a paid LINBIT Starter plan at $1,000 a year; consider it when evaluating a separate storage replication option.
• XtreemFS is a free, BSD-licensed open-source alternative for Linux, macOS, and Windows self-hosting.
• OpenEBS is a free, open-source Kubernetes storage option for teams whose focus is Kubernetes storage.
• Amazon S3 is a hosted, paid alternative with usage-based billing and no minimum charge; consider it if you want a managed service rather than operating a self-hosted server.
• Longhorn is free, open-source distributed block storage for Kubernetes, rather than an S3-compatible object store.
• Resilio Active Everywhere is a paid alternative with a free trial and pricing based on devices or endpoints rather than data volume or transfer usage.

Verdict

Choose Obstor if you want self-hosted S3 compatibility with native SFTP and can operate a distributed deployment. Its combination of familiar access methods and replication features is the central reason to choose it. Look elsewhere if you need a standalone production setup with versioning, object locking, or bucket replication, or if you prefer not to manage storage infrastructure yourself.