We often get called in to fix issues after Tableau server has been deployed. At this point, our clients are experiencing issues with slow performance, low user adoption and data governance among other things. There are many aspects to consider before rolling Tableau server out company wide. Below, we list the most common pitfalls we have come across in our work with clients and tips on how to avoid them.
Learn more details on pitfalls in our on-demand webinar: The Perks and Pitfalls of Tableau Server.
- Under powered hardware. Tableau is interactive, and this puts a big load on the server. Under spec’ing your hardware will just cost you more in time and money down the road.
- Installing with Tableau defaults. Settings are not easy to change after software is installed, and they can have a huge impact on scale. For instance, Tableau only allows for one authentication provider to be supported and it can’t be changed post installation. So, if you decide to do active directly vs local and change your mind, you have to do a complete reinstall. You’ll also want to consider cluster configuration upfront so that you won’t have to reconfigure down the road.
- Insufficient care and feeding of the server administration environment. Tableau’s admin environment builds in many useful capabilities that, when utilized, can enhance performance and provide a better user experience. For example, the database views function lets you see what data sources are most frequently being used and which aren’t to then decide if and where you can free up space. Data source rationalization should be a part of your regular care and Tableau does lots to help this.
- Lack of governance. This aspect goes beyond the administrative side to take into account the combination of people and processes that need to accompany the technology. Tableau server is great, but its technology alone won’t give you a good environment. One of Tableau’s greatest strengths can also be its weakness: with Tableau, metadata exists at the workbook level and is not centralized as with enterprise BI systems like Cognos. Since Tableau users can create metadata several times over, you need good governance in place.
- Out-of-date Tableau versions. Because Tableau has introduced significant performance enhancements with each major version, it’s really important to stay current. Big changes occurred in particular between versions 8 and 9, which saw improvements of 2-50x in overall performance. One of our clients saw high double digit decrease in load times for nearly all workbooks simply by upgrading from 8 to 9!
- On the other hand, don’t just jump to the latest version assuming it will fix everything. Do your homework, look for known issues, maybe have a few others test the waters first, and have a testing plan that represents your use cases. And ALWAYS perform backups so you can fall back should you encounter the unexpected.
- Inadequate data prep. If we were to place in order the most common complaints we hear from clients, this would easily top the list. Good data prep is critical to automating and building out high performance business analytics, period. (And it’s also an area of expertise where we at Senturus really shine, if we do say so ourselves). Good data prep is critical with dashboard designs where you’ll be pulling in a combination of data sources and want to provide maximum performance, security and user experience. Whether we’re talking about data transactional systems like CRMs, marketing systems like Marketo or Eloqua or ERPs—none of these source data structures can just be plugged into Tableau and expected to perform. There are many considerations to take into account, such as: What type of database do you use? Extracts, yes or no? What will be the implications of Hyper (Tableau’s next gen data engine) when it gets rolled out? What kinds of ETL/data prep should you perform? There are a dizzying number of tools out there.
- Winging it with security. Not putting adequate consideration into your security (e.g. site permission rules, content ownership, group permissions and the like) will cost you in many ways. We could do an entire blog on this one subject alone. There are many different levels at which to apply security. As a rule, the higher the level you can apply, the less maintenance. But again, there are lots of different security components, and if they are not understood and applied correctly, you can end up with unpredictable behaviors.
- Always consider end users. Taking into account end user needs doesn’t stop with how you deploy server; it extends to dashboard design. Designing for the end user experience will increase user adoption. You’ll want to consider, among other things, what device the dashboard will be viewed on, such as a high-resolution monitor, a tablet or an iPhone. While Tableau helps the process by offering dashboard specific layouts, you don’t want to simply rely on Tableau to auto-fit your larger screen layout to a smaller device. What translates great on the big screen gets smooshed up and unreadable on a small screen. Take some time to pare down information and make to fit correctly. We discussed this subject further and did a demo in our on-demand webinar Rolling Out Tableau to the Enterprise.
- Insufficient training. Among the biggest barriers to entry we hear from our clients is insufficient training. While BI power users may easily make the shift from one tool to another, don’t assume all your end users will make the transition to server as easily. Train them to understand the differences…and the capabilities, enable them to understand the data they are working with. Encourage them to ask questions. Similarly, from an administrative perspective, the team needs training to understand the admin functionalities, config utility, tabadmin and more.
Check out our on-demand webinar The Perks and Pitfalls of Tableau Server, to learn details on all these points…and to learn the 11th pitfall. Yes, there’s a bonus pitfall: we just couldn’t limit ourselves to 10.
