Tableau Server Best Practices for Developers

Tableau Desktop users can utilize the Tableau Server provided by Pitt IT to share dashboards. The following are practices and recommendations for utilizing the University of Pittsburgh’s Tableau Server, located at https://analytics.pitt.edu

Accounts and Projects

User Accounts

All faculty and staff users at Pitt can utilize the Tableau Server. To log into the Tableau Server, a person must utilize a University of Pittsburgh account, including sponsored accounts. Users outside of the University can only interact with the Tableau server through a public dashboard offering (see below).


Groups

You should plan to utilize CDS groups whenever possible for governing access to Tableau dashboards. To create a CDS group, you’ll need to work with your RC security contact. Once you have created a CDS group, contact the Pitt IT Help desk and request that the group be added to the Tableau Server. The Pitt IT analytics team will add your group to the server’s sync list, and you can now attach your group to Tableau dashboard and data sources for access control.


Projects

You can request the creation, deletion or changes to a Tableau Project folder by opening a Helpdesk request. Please provide the name of the requested group - the Pitt IT Analytics team member implementing your request may work with you to align your request along existing naming conventions. As the requestor, you will be made the owner of the Project folder, and can assign permissions as you desire. For beginner users: we recommend you set and lock permissions https://help.tableau.com/current/server/en-us/permissions_projects.htm#LockProjectto make sure that access is granted in a consistent manner.

For advanced users: you might let each author / publisher manage permissions to their own asset. If you use this approach, people are likely to see unusual behavior (e.g., users may be able to see the workbook but they can’t see the project it belongs to; or users can see one sheet but not all sheets within a workbook.

As the owner of a project, you may also create sub-project folders.

Public Dashboards

Tableau dashboards can be configured to be visible to the general public. These dashboards can be embedded in web pages, or shared with those outside of the University. To make a dashboard public, please contact the Technology Help Desk with details about your request. A member of the Pitt IT Analytics team will work with you to review your dashboard (to ensure no PII or sensitive data is being released) and publish it for public consumption.

Examples include…


Data Sources

The Pitt IT team recommends the use of Published Data Sources, especially for data that needs to be reused in more than one dashboard. See https://help.tableau.com/current/pro/desktop/en-us/publish_datasources_about.htm#what-makes-up-a-published-data-sourcefor more information.

Selecting the right approach - Live Connection vs. Extract

Tableau supports multiple ways of getting data to your dashboard. Selecting your approach carefully will ensure that your dashboard works smoothly and responsively for your users.

Tableau live connections to databases allow Tableau to generate SQL queries and execute them against your database. This technique allows your database to do the data retrieval and processing, rather than Tableau. It is a great choice in a number of use cases, but does require you as a developer to have some visibility and ability to work with the database in question.

Tableau Extracts are a powerful feature that allows the Tableau server to pull data from a database or other source, and store it in a fashion optimized for Tableau dashboards. While extracts are powerful, there are important factors to consider…

  • How frequently is your data refreshed? Do not schedule a Tableau extract to run more frequently than the data is refreshed - this will minimize Tableau Server delays.
  • How frequently do your users need the data refreshed? While your underlying data may be refreshed daily, users may only consume the report/analysis at a weekly level. When possible, set the refresh of a Tableau extract to run only as frequently as it will bring value to the users of your dashboard.
  • If your data accrues slowly and includes timestamps, consider implementing incremental extracts. Incremental Extracts use a date field to only bring in new records to your Tableau data source or dashboard.

Selecting a Schedule for Extracts

Things to keep in mind when selecting a schedule for your Tableau Extract…

  • The Tableau Server is busiest between 6:00 a.m. and 9:00 a.m., as data warehouse and other data refreshes complete and Tableau extracts begin. If possible, choose a time outside of that window to help ensure your job completes when schedule.
  • Adjust your refresh schedule to ensure that each refresh is providing value to your customer.

Do not refresh the data more frequently than it updates in the source. Scheduling frequent extract refreshes does simply causes more delays for yourself and others in the broader University community. In addition, if you’re connecting to a database, instantiating multiple refreshes connections will create drag on your database and could cause a lock on your account.

Subscriptions

Subscriptions run as part of the same work queue as extracts. Where possible, guide users to scheduling subscriptions outside of the 6:00 a.m. - 9:00 a.m. busy hours to ensure subscription messages arrive on time.

If you would like to request a new schedule frequency for subscriptions, please contact the Pitt IT Technolgoy Help Desk and request the creation of a new Tableau sched