Tuesdays with Tableau: Why I Love Tableau Server

By Gail Kluepfel

These days, who doesn’t want to have a self-service business intelligence platform? Business users benefit from generating reports without having to wait for an IT resource or centralized reporting team to create them. IT teams benefit by being able to focus on the heavy lifting of developing and integrating data across core systems rather than getting randomized by users requesting reports. But “self-service” doesn’t translate to “hands-off” for IT – someone needs to refresh the data.

A case in point: I recently caught up with a client for whom we implemented a SharePoint-based, self-service Power BI solution. Three years in, the system was still working well – almost too well! His only complaint was that his primary business user had gone wild building Power Pivot models in Excel, rather than taking the time to work with IT to deploy the changes to the initial model we built on their Analysis Server as a tabular cube. Now their IT director reports a profusion of workbook models built off the original that must each be refreshed separately, with multiple jobs for the IT team to babysit.

Is there no end to such madness? Happily making the switch to using Tableau desktop, many of our current clients are delighted to find that Tableau Server provides a more efficient way for IT to help manage self-service BI. While much has been made of Tableau as a great visualization tool (it truly is!), I want to give equal time to how Tableau Server handles data refreshes. In short, Tableau Server efficiently manages access to data sources and provides a collaborative platform for reporting, but it also reduces the burden on data sources and IT by managing refresh automation.

How is Tableau Server more efficient when refreshing data?

In self-service tools like Excel, the data models built by business users (each with unique calculations, measures for grouping, and renaming of fields for specific business purposes) reside in many different workbooks, even though they all may be hitting the same data sources. By contrast, Tableau’s data model design is based on querying data sources at run time. This means that even when users across different functional areas build vastly different data models, and save them to different workbooks, the shared data sources can still be refreshed with a single job.

The architecture of Tableau makes it a much better platform to enable business self-service reporting without the headache of supporting refreshes of hundreds or thousands of variations on the same data sources.

Tableau's tiered architecture includes a super-fast data engine that enables direct queries to data sources, along with a Data Server component to store extracted data in reports. (click image to view full size)

Consider Sue, a user in Marketing. She creates a workbook connecting to a CRM database serving sales data and a web metrics database serving web traffic data. She creates custom measures and aggregations appropriate to her business reporting needs, then posts this to a SharePoint site, where her IT team has enabled a daily refresh of her underlying data. Sue’s colleague Tom, in Sales, connects to the same CRM database and creates a sales pipeline and forecasting report, posting a separate workbook on the same SharePoint site. Now IT has to run another refresh job.

This scenario is very different with Tableau Server. Whenever a user saves a workbook to Tableau Server, the data connections and extracts become part of the shared repository of data sources, and any workbook connecting to a given data source updates simultaneously. This is much easier for IT teams to manage than hundreds of individual workbooks with separate refresh jobs.

Tableau Server is a Data Server

Another benefit of extracting data from source systems is that individual users neither need access to nor query source systems directly. Tableau Server can extract the report data on an automated schedule, so that users only need to have permissions to the data on Tableau Server. And, because one data source extract can be used by many workbooks, you save on server space and processing time. As a recent Tableau white paper explains, “When a workbook using a Tableau Server data source is downloaded, the data extract stays on the server, resulting in less network traffic. Finally, if a database driver is required for a connection, you only have to install the driver once, on Tableau Server, instead of multiple times, on all your users’ desktops.”

Tableau Server supports all the self-service reporting capabilities business users could want, alongside the efficient data governance that IT teams need to prevent a “wild west” of competing data models and data refresh jobs. With such great features, Tableau Server really makes Tableau Desktop a great enterprise business intelligence platform, and not just a great visualization tool.