Skip to content
Tutorials Center
My learning

Topics

Lesson 24 of 35 ยท Sharing and Security

Publishing to Tableau Cloud and Tableau Server

Publish workbooks and data sources to Tableau Cloud or Server: projects, embedded vs published data, refresh schedules, Bridge, and a checklist you can run.

A workbook on your laptop helps one person. To share it, you publish it to a Tableau site, where colleagues open it in a browser, on a phone, or inside other applications. The site is either Tableau Cloud (hosted by Salesforce, formerly Tableau Online) or Tableau Server (installed and run by your own team). The publishing experience is nearly identical; the main difference is who maintains the machines.

Tableau cannot run in this page, so the lesson explains the steps in the application. The runnable code is a small stand-in: pandas scripts that compute refresh schedules and run a pre-publish checklist, the kind of planning you do around a real publish.

The architecture in one picture

flowchart LR
  A["Tableau Desktop<br/>(author)"] -->|"Publish"| B["Tableau Cloud / Server<br/>(site)"]
  C["Databases, files,<br/>web services"] -->|"live query or extract refresh"| B
  B -->|"browser, mobile, embed"| D["Viewers"]
  E["Tableau Bridge<br/>(private network)"] -.->|"refresh on-premises data"| B

Authors publish to a site; the site talks to the data; viewers see the result in a browser.

Publishing a workbook

In Tableau Desktop choose Server > Sign In, enter the site address (for Cloud, something like a pod URL plus your site name) and sign in. Then Server > Publish Workbook. The dialog asks you:

  • Project: the folder on the site where the workbook goes. Projects hold permissions (see Permissions and Row-Level Security).
  • Name and description.
  • Sheets: which sheets and dashboards to publish. A sheet used only inside a dashboard can be hidden from navigation.
  • Data Sources: how the connection is stored (see below).
  • Permissions: usually inherit from the project.
  • Show sheets as tabs: whether viewers see a tab bar.
  • Edit credentials for data access.

After publishing, a browser window opens the new view. Viewers see the workbook through a URL and, depending on their licence, may edit it in the web (Explorer and Creator licences) or only view and interact with it (Viewer).

Embedded or published data source?

This is the most important choice, and it comes up for every workbook.

  • Embedded: the connection and the extract live inside the workbook. Simple, but ten workbooks on the same data mean ten extracts, ten schedules and ten places to fix when a column changes.
  • Published data source: the connection or extract is published once, separately, and workbooks connect to it. There is one refresh, one set of calculated fields, and one set of permissions. Governance teams certify and recommend these.

Embedded data

Inside the workbook Simple and self-contained.
Copies pile up Every workbook refreshes alone.
1 / 2

To publish a data source, right-click it in the Data pane in Desktop and choose Publish to Server (or on the Server menu). In the same dialog choose the authentication (see credentials below) and whether to Include external files for file-based data.

Live or extract on the server

A live connection queries the database every time somebody opens or interacts with a view, so you always see current data but you load the database. An extract (a .hyper snapshot) lives on the site, is fast, and is refreshed on a schedule. The next lessons go deeper (see Extracts and Performance Tuning). When you publish an extract you pick:

  • Full refresh: rebuild the extract from scratch.
  • Incremental refresh: add only rows newer than a date or ID column.
  • Schedule: for example every day at 06:00, or every hour. Schedules are defined by an administrator on Server, and on Cloud there is a similar set, plus refresh tasks you can attach per extract.

Credentials and Bridge

For live or refreshing connections the site has to log into the database. Options:

  • Embedded password: the site stores a service account's credentials. Easy; all viewers see the same data.
  • Prompt user: each viewer is asked for their own credentials. Good for database-level security, more friction.
  • Viewer credentials / SSO / impersonation: pass identity through to the database.

If the data is in a private network (an on-premises SQL Server, a file share), Tableau Cloud cannot reach it directly. Tableau Bridge is a small client installed inside your network that keeps a secure outbound connection to Cloud and handles live queries and extract refreshes. For Tableau Server you normally deploy the server inside the network, so Bridge is not needed.

Local file-based sources (Excel and CSV on your laptop) do not refresh once published, unless the file lives on a network location the site can reach or you set up Bridge or a file-sync process. A common surprise is a dashboard that stays frozen at the day somebody published it.

Tableau Cloud vs Tableau Server

Tableau Cloud Tableau Server
Hosting Salesforce Your servers or your cloud account
Upgrades Automatic You schedule them
Network Needs Bridge for private data Sits inside your network
Control Less (managed service) Full (authentication, scaling, nodes)
Cost model Subscription per user Licences plus infrastructure

Choose Cloud when you want no servers to maintain and your data is in the cloud or reachable with Bridge. Choose Server for strict data residency, heavy customisation, or air-gapped environments.

Publishing hygiene

Before you click Publish:

  1. Name things well. Sales Dashboard v7 FINAL copy does not scale.
  2. Remove unused data sources and sheets. Smaller workbooks load faster.
  3. Set the starting view and sizes. Fixed-size dashboards render faster than automatic ones.
  4. Test with a normal viewer account (not an admin) to see what people really get.
  5. Add a description with owner, data source and refresh time.

You can run a boring but valuable checklist like this one in code. The stand-in below holds a list of workbooks (as an admin might export them from a site inventory) and flags problems.

Python Pre-publish checklist

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

And a quick look at what a schedule means in practice: how many refreshes a day an hourly schedule between 06:00 and 18:00 creates, and when the next ones run.

Python What does a schedule produce?

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Hourly runs mean the data on a dashboard can be up to an hour old, plus the time the refresh takes. Say so on the dashboard ("Data refreshed 10:02") so nobody mistakes it for real time.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Ten workbooks use the same sales data. What is the best setup?

    A published data source gives one refresh, one set of calculations and one place to apply security.
  2. 2. Your data sits on an on-premises SQL Server and you use Tableau Cloud. What do you need?

    Bridge runs inside your network and lets Tableau Cloud reach private data for live queries and extract refreshes.
  3. 3. A published workbook uses an Excel file on your laptop. Next week the dashboard shows old numbers. Why?

    Local files cannot be refreshed by the site unless they live on a reachable location or you use Bridge.
  4. 4. Which refresh type adds only new rows based on a date column?

    An incremental refresh appends rows newer than the last value of a chosen column; a full refresh rebuilds everything.

Summary

  • Publish workbooks and data sources from Desktop (Server > Publish) to a project on Tableau Cloud or Server.
  • Prefer published data sources for shared data; keep embedded sources for one-off work.
  • Extracts refresh on schedules (full or incremental); live connections query on demand.
  • Use Bridge for private data on Cloud; local files do not refresh by themselves.
  • Check names, descriptions, sizes and freshness before publishing; test as a viewer.

Next: Permissions and Row-Level Security