Monitoring your Tableau Server
Recently I participated in my first virtual conference, the Tableau Fringe Festival EMEA, presenting on a topic close to my heart. Administering Tableau server. My talk was more about sharing some of the learnings that I have discovered over the past 18 months. You can view a recording here.
The Tableau community is fantastic. I’ve never experienced anything like this before in terms of people sharing and collaborating knowledge. It is so much more than just sharing how to do things with the software. On top of some incredible product tips and tricks, people share their ways of working and thinking about data visualisation . Many other software companies I have used in the past restrict this knowledge from being publicly available and try to force people into purchasing training courses.
This concept is so outdated and static. I felt alienated and alone. I ended up resenting the product I was using because it was so difficult to learn and improve my skills and knowledge. Tableau enables this learning and has created more than just a knowledge repository, they have created a movement of data lovers.
There is so much content available on Tableau desktop and as someone who spends his time working with Tableau server I felt there is a bias towards desktop. My talk was aimed at informing and educating other Tableau server admins, especially those who are new to this role or find themselves performing server admin duties when previously just desktop users.
One of the areas I covered in my talk was around monitoring Tableau server. I’m creating this post to share some of the workbooks that I have created which I use to monitor user behaviour on Tableau server.
When I first started looking after Tableau server I found that there were lots of dashboards already created. The problem is I didn’t create them and although they were useful, they were not able to answer my specific questions.
My questions about server activity were around “Is my server running well?”, “Can people do what they want to?” and “Are people using the tool properly?”.
Based on these questions I embarked on a project to create a “command centre” of dashboards using the PostgreSQL datasource connecting you to all kinds of data from the Tableau server repository.
- No one uses Tableau
I was constantly in meetings where people would question how many users were accessing the server. Often there was suspicion or doubt when I would rattle off the current stats. I decided to end this by creating my first command centre dashboard which gave a view of unique users for the past six months. Supporting this headline figure was concurrent users, views, views per user, published workbooks and subscriptions sent. Credit to Mark Jackson from his blog post of creating custom server admin dashboards.
2. Extract run times
Sometimes I feel for our server. It takes a lot of abuse, especially in the form of data extracts. As much as I try to protect the server, some extracts really chew up a lot of time and processing power from the server and the source database. I created a view of extract run times to expose the data extracts which are just plain crazy. It gives me sight of poorly designed dashboards which allows me to identify training needs for the users publishing these type of dashboards. I can see the wait times in orange blocks of time and the run time in green. I reworked an original version of this workbook from this post by Tableau Product Manager — Michael Chen.
3. What are users viewing on Tableau?
This dashboard breaks down groups of users based on the number of views and the number of workbooks they are viewing. This really helps me understand user behaviour. Are people just viewing one popular dashboard over and over again? Or are users exploring other content that is being published?
4. What content is not being used?
One of my biggest frustrations is users who publish content that is looked at a handful of times and then they just leave it there. It sits there gathering dust but normally has a data extract refreshing each day. My thoughts are that if you have the permission to publish you also have the responsibility to remove old content when it is no longer being used. This dashboard gives me a view of workbooks which are no longer being viewed and I can check if there are any data extracts which can be deleted.
If you have access to the Tableau PostgreSQL you can really find some valuable tables of data to analyse. On top of the examples I’ve listed above I’ve also used the PostgreSQL tables to find out which desktop licences have a different registered email and assigned emails.
Every Tableau server admin should have their suite of server reporting to help understand how their users are interacting with Tableau server. I’ve saved these workbooks in a dropbox folder if anyone wants to look and use these to help shape their own server monitoring. Just remember you need to enable your PostgreSQL and edit the data connection for your own server.
Special thanks to Joanna Hemingway who helped create some of these views when working at Just Eat.
