Skip to main content
To homepage To homepage

Hi all,


We recently started developing a new data platform using TimeXtender with Tableau as a visualization tool. We are currently determining how to setup our endpoint in TimeXtender for Tableau as we have several options, each with their pro's and con's. Im wondering what choice others made and if maybe there's something we haven't thought about yet while trying to make this choice.


As we see it now we have three main options:


- Multiple .tds files, one for each fact table. As Tableau requires all tables to have a relation with eachother it is not possible to have a .tds file endpoint with all facts as not all facts relate to all dimensions. Pro's: smaller, efficient sources with good manageable security. Cons: Can become difficult to maintain with lots of .tds which share dimensions, for example when an attribute is added to a dimension this has to be done multiple times. When using multiple .tds sources in a Tableau sheet there's no way to link shared dimensions and thus fltering becomes difficult and unclear to users.


- Single .tds file for all facts and dimensions using a linktable with all possible dimension key combinations from all fact tables. Using a link table we can ovecome the issue where we cannot have all fact tables in one endpoint. Pro's: Single endpoint so easy to maintain. Cons: We expect performance to become an issue when the link table gets too large, unclear for dashboard builders what dimensions can be used with which facts.


- SSAS tabular model, as it is possible with Tableau desktop to connect with a tabular model we also want to look at this option. Pro's: one endpoint so efficient to maintain, makes us more flexible with dashboarding tool because a possible switch to Power BI will be easy as custom measures will already be in DAX. Cons: Not possible to connect via Tableau server, meaning builders will have to download datasource to desktop, build dashboard and republish, not the most efficient way. 


Our own preference with above pro's and cons in mind is using a tabular model, but we do find it a big con that this will not work with Tableau server. However, the cons of the first two options using .tds files seem to have a higher impact as this can become more and more difficult as the data warehouse grows, whereas the con for a tabular model will stay the same regardless of model size. 


We're curious if there are others having made the same decision and if our pro and con list is complete. Looking forward to any input, thanks in advance.


 


Bart

  • I would definitely not recommend a single .tds file. Because of how tableau consolidates the data into a single table or view, this could dramatically explode the amount of rows and performance would be terrible. 

  • If using Tabular models it would be recommended to have multiple data marts and would need to update multiple models for conformed dimensions, however the amount of time it takes to drag a field from a table to a few semantic models is negligible. This would take a fraction of the time that it would take to update the tableau dashboards. 

  • So, if using Tableau server is important to you. Multiple .tds files would be the recommended approach. 

Hope this helps!


 

Hi Joseph,


Thank you for your response. Good to know that the first option is not a good choice due to performance. Tableau server is important for us, as that's the way dashboards are shared in the organisation. Therefore the multiple .tds files remains as only / most viable option. We will use this and see if we can mitigate impact of maintaining and pushing multiple files between servers.


Thanks,


Bart

Reply


Terms and ConditionsCookie settings