Storage Backends

What this section covers

Tutorials here are organized by backend type: how to prepare the external service, create the storage policy, configure policy group rules, move users or teams over, and validate before going live. The two-layer concept of storage policies and policy groups itself lives in Storage Policies and Policy Groups.

Choosing a Backend

Backend	Connector ID	Deployment scope	Best for	Tutorial
Local	asterdrive.storage.local	Deployment-managed	Single machine, NAS, small teams, minimal dependencies	Local
S3	asterdrive.storage.s3	Shared across Primary instances	S3-compatible object storage, external buckets, and large files	S3
Alibaba Cloud OSS	asterdrive.storage.alibaba_oss	Shared across Primary instances	Alibaba Cloud OSS with native V4 signing, split endpoints, or CNAME	Alibaba Cloud OSS
SFTP	asterdrive.storage.sftp	Shared across Primary instances	SSH/SFTP file servers and server-side streaming	SFTP
Azure Blob	asterdrive.storage.azure_blob	Shared across Primary instances	Azure Storage accounts and Blob containers	Azure Blob
Huawei Cloud OBS	asterdrive.storage.huawei_obs	Shared across Primary instances	Huawei Cloud OBS with native OBS signing and regional or custom-domain endpoints	Huawei Cloud OBS
Tencent COS	asterdrive.storage.tencent_cos	Shared across Primary instances	Tencent COS and per-policy COS CI processing	Tencent COS
Remote	asterdrive.storage.remote	Shared across Primary instances	Objects stored by another AsterDrive follower node	Remote
OneDrive	asterdrive.storage.onedrive	Shared across Primary instances	Microsoft 365, OneDrive, SharePoint, and group drives	OneDrive
Qiniu Kodo	asterdrive.storage.qiniu	Shared across Primary instances	Qiniu Cloud Kodo S3 spaces with official endpoint diagnostics	Qiniu Kodo

For multi-Primary (cluster profile) deployments, the default policy must be reachable by every Primary. local may be the default when the operator shares both the policy path and upload_temp_dir; see Storage Policies and Policy Groups.

For each backend’s direct-upload capability, capacity observation, native processing, and credential mode — plus how to choose between relay_stream and presigned — see the Storage Capability Matrix.

Shared Onboarding Flow

Do Not Switch Production Traffic Yet

For a new backend, create a separate new policy; do not edit the old policy that is in use.

Recommended approach:

1. Create the backend policy
2. Create a test policy group
3. Bind a test user or test team
4. Run upload, download, share, delete, and restore end to end
5. Only then migrate real users or teams to the new policy group

Do not edit the real landing location of a policy that already has files

The local directory, S3 / OSS bucket / endpoint / prefix, Azure Blob endpoint / container / base path, OneDrive drive / root item / site / group targeting fields, SFTP endpoint / base path, and the bound remote node all decide where old files live. Change them in place and old files may become unfindable. For the correct move procedure, see Storage Policies and Policy Groups.