canot

A departure rehearsal, in your accounts.

Your administrator sets up read-only access, then one command runs everything. You do not leave your provider, and you have nothing to request from them.

Everything happens in your accounts: your data platform, and the host you choose
  1. Set upA small open warehouse is created at the host you chose: one machine, one storage bucket, one catalog of Iceberg tables.
  2. CopyYour data is read with no ability to modify it, then written in Iceberg format to another of your accounts.
  3. ReplayThe 20 most important queries, chosen with you, run on both sides.
  4. CompareTables and results are compared (row counts, per-column statistics, content fingerprint), with tolerances written in advance.
  5. SignAn evidence file is produced: parity, duration, cost, and what does not move as is.

Canot sends nothing to Q4O: no telemetry, no server.

  • canot scanFree read-only X-ray: volumes, objects that do not move on their own, estimated exit duration and cost.
  • canot drillThe full drill and the evidence file.
  • canot teardownDeletes what was copied.

Platforms.

The scan and the exit drill work today on Snowflake and BigQuery. Others are added with a design partner: the connector is written and validated with you, on your account.

Available

Snowflake and BigQuery: read-only scan, copy, comparison, query replay, signed file. dbt: scan of the project files and replay of the models on the copy.

On request

Databricks, Microsoft Fabric and Azure Synapse, Amazon Redshift, Teradata, Oracle, SAP Datasphere, ClickHouse. Tell us yours.

Beyond warehouses

Analytics and data science tools (Dataiku, SAS, Alteryx), semantic layers of BI tools: an inventory of what does not move as is, before any copy drill.

An "on request" platform is not yet validated on a real account: we tell you so before you commit. Write to us.

Exit locally, to your own infrastructure.

No cloud needed at the destination: a folder is enough. Canot writes the copy in Apache Iceberg format to your disk, your NAS or an S3 storage you host, with the evidence file next to it.

uvx canot drill "snowflake://MY_ACCOUNT/MY_DATABASE?user=MY_USER" /data/iceberg
  • /path/to/folderLocal disk, network share or mounted NAS.
  • s3://bucket/canot?endpoint=https://s3.internal:9000Your S3-compatible storage (MinIO, Ceph, NetApp…), with an internal proxy and certificate authority if needed.
  • --catalog env:CANOT_CATALOG_URIA PostgreSQL catalog of your own; the password stays in an environment variable.

The copy does not depend on Canot: DuckDB, Spark and Trino read it, and the reread.sql file recounts every table without the tool. Details are in the documentation.

You choose the host.

Canot writes to any S3-compatible storage. Shortcuts exist for French and European hosts.

  • ScalewayFrance · Netherlands · Poland
  • OVHcloudFrance
  • OutscaleFrance
  • IONOSGermany
  • STACKITGermany
  • AWSParis and other regions

SecNumCloud qualification, when it matters to you, depends on the host's specific offering: check the ANSSI list. The Data Act is the same across the European Union.

A question, or a test to suggest? Write to us