Video | Tableau | Data prep | Tool strategy | Analytics

Publishing data sources to Tableau Server & Tableau Online: Tableau Tutorial for Beginners:

If you want to share a data source properly, you need to publish it to the server, and there are a few quirks worth knowing before you do.

Part ofTableau Server
Watch on YouTube
  • Publishing a data source centralises database access, governance and driver management, lets teams collaborate on a shared source and adds mobility through web authoring.
  • A published live connection saves connection details (a TDS) to the server, so every interaction calls the database, while an extract is a snapshot that needs a refresh schedule set up on the server.
  • Authentication can either prompt each user for credentials or embed a password, and only extracts can be set to allow refreshes (which is where the server refresh schedule comes from).
  • Overwriting a published data source wipes any field-level metadata or descriptions you previously added on the server, so re-apply them after each publish.
  • You can switch a workbook between a published source and a local copy using 'create local copy' and 'replace data source', but taking extracts of extracts quickly makes it hard to track what you're actually connected to.

Publishing a data source to Tableau Server or Online centralises database access and governance, lets teams collaborate on one shared source, and adds web-editing mobility — but live and extract connections behave differently once published, and there are a few traps around metadata, authentication and nested extracts worth knowing before you do it.

Tim works in Tableau Desktop with a test workbook connected to two data sources hitting the same Snowflake database — one set up as a live connection, one as an extract — to show how publishing each behaves differently.

The Breakdown
  1. Why publish a data source 1:56

    Publishing centralises database access and driver/credential management, lets people collaborate on a shared governed source, and adds mobility since published sources can be used in web authoring from any machine.

  2. Prep the live connection before publishing 4:14

    Before publishing, Tim renames a poorly-named field and splits a compound ID field for clarity, since this is the point you'd typically validate and clean up the data structure.

  3. Publish to Tableau Online 5:48

    Right-click the connection and choose 'Publish to Server'. He explains that this pushes the TDS (connection details plus any metadata) up to the server, then walks through project selection, name, description and tags.

  4. Permissions and authentication 8:36

    Permissions inherit from the project by default and can be edited via presets, though locked project permissions override anything set here. Authentication can either prompt each user for credentials or embed a password — live connections don't get the extract-refresh option.

  5. Live vs published behaviour 15:58

    Once published, users can't tell whether a source is live or extract from the front end — every interaction on a live source calls the database directly, which could be used to gauge usage.

  6. Republish and switch authentication to prompt 16:51

    Republishing over an existing data source overwrites it; changing authentication to 'prompt the user' means each person logs in with their own credentials. Republishing wipes any field descriptions added on the server, so they need re-adding.

  7. Create local copy and switch back 19:23

    'Create local copy' saves a TDS/TDSX version of the published source locally; 'replace data source' then swaps the workbook onto it. Useful for testing locally before switching back to the published version, but risky to juggle for long.

  8. Publish an extract, set refresh, certify 21:12

    Publishing an extract offers 'allow refresh' instead of embedding a password, and a Tableau Bridge warning if the source is behind a firewall; refresh schedules are then set up on the server. He also shows certifying a data source (adding a description/approval badge) and warns that taking an extract of an already-published extract creates a confusing chain that's easy to lose track of.

Worth Knowing
  • Overwriting a published data source on republish wipes any field descriptions or metadata added on the server — you must re-add them each time.
  • Only extract connections offer the 'allow refresh' authentication option; live connections can only prompt for credentials or embed a password.
  • Taking a local copy of an already-published extract (which was itself built from a live connection) creates a tangled chain — right-click properties to check what you're actually connected to.
  • Extracts behind a firewall need Tableau Bridge to refresh, since the server can't reach the database directly over the internet.
Use It When

Reach for this when you need to share a database connection with a team without distributing credentials, want a governed single source of truth others can build on, or need extracts refreshing on a schedule rather than manually in Desktop.

How this Rollup was made provenance & method

A Rollup is drafted by AI from the video's transcript, then reviewed and edited by Tim. Everything used to produce this one is listed below — the model, the exact prompt, and the source video — so the process is transparent and reproducible.

Transcription
On-device — NVIDIA Parakeet v3 for recent videos, OpenAI Whisper large-v3 for earlier ones. The transcript never leaves the machine or gets published.
Drafting
Claude Sonnet 5 in the cloud, from that transcript.
Prompt
The exact Rollup prompt (v1) — the full system prompt, unedited.
Source video
Watch on YouTube
Drafted
4 July 2026 at 16:06
Reviewed & edited
5 July 2026 at 09:41 · by Tim Ngwena

Model + prompt + video is everything you'd need to recreate a Rollup like this yourself. The one thing we don't share is the transcript.

Rights. The video and its transcript are the property of TN Media Ltd. Unauthorised use or download is prohibited. © TN Media Ltd.