For teams needing point-in-time data recovery

    Recreate any dataset,
    from any point.

    S3-compatible storage with built-in version history. Restore any prior state by version ID. $4.99/TB flat, no egress on reads.

    Start for freeTalk to an expert

    No credit card required · No egress fees · Connects in minutes

    A backup is not the exact prior state.

    Scheduled backups capture moments in time, not the continuous history. Debugging a data-quality regression, reconstructing a prior model training state, or answering an audit request often requires a specific state between backups.

    The backup gap

    Backups are periodic. Bugs are not.

    A nightly backup captures yesterday. If a data-quality issue was introduced at 2pm and discovered at 11pm, the prior-clean state is between two backup windows — and unavailable.

    The debug state

    Reproducing the exact input requires the exact prior state.

    A model trained on a dataset that was later modified cannot be reproduced without the exact version of the training data that was used. Version history makes that retrieval a one-line query.

    The audit requirement

    Auditors ask for the specific record, not a backup.

    Evidence requests for specific records at specific points in time require version history. A backup restore retrieves the whole snapshot — version retrieval returns the specific object at the specific time.

    Standard S3 versioning. No egress on reads.

    Enable versioning on a bucket. Every subsequent PUT stores a new version. Retrieve any prior version by ID — the API is the same S3 call.

    versioning.py
    import boto3, os

s3 = boto3.client(
    "s3",
    endpoint_url="https://eu-west-1.s3.fil.one",
    aws_access_key_id=os.environ["FIL_ACCESS_KEY"],
    aws_secret_access_key=os.environ["FIL_SECRET_KEY"],
    region_name="eu-west-1",
)

# Enable versioning on the bucket
s3.put_bucket_versioning(
    Bucket="dataset-store",
    VersioningConfiguration={"Status": "Enabled"},
)

# List all versions of a key
versions = s3.list_object_versions(
    Bucket="dataset-store",
    Prefix="training/labels.parquet",
)

# Retrieve a specific prior version — no egress charge
obj = s3.get_object(
    Bucket="dataset-store",
    Key="training/labels.parquet",
    VersionId="v_20240915_abc123",
)
prior_state = obj["Body"].read()

    Version history that costs what it stores.

    Standard S3 versioning API. Versions billed at $4.99/TB like any stored object — no retrieval fee on any read.

    S3 version history

    Every PUT creates a new version. Prior versions are retained and retrievable by version ID using standard S3 GET. No extra configuration.

    No egress on version reads

    Retrieving any prior version — for debugging, audit, or reconstruction — is included in flat storage. No egress charge per read.

    Integrity verification

    Every stored version is verified approximately every 24 hours. Prior states remain verifiably intact, not merely retained.

    Flat cost across versions

    Versions cost $4.99/TB like any stored object. The version history grows with the data; the rate per TB does not change.

    One rate. $4.99/TB/month.

    Storage — versions included. No retrieval fee, no egress per read. The whole version history costs what it takes to store it.

    Start for freeTalk to an expert

    No credit card required · No egress fees · Connects in minutes

    Not a backup. The exact state.

    Free 1 TB evaluation. Enable versioning, write a few objects, and retrieve a specific prior state in one call.

    Start for freeTalk to an expert

    No credit card required · No egress fees · Connects in minutes