Documentation
-
If you receive the following message while attempting to install VIZYUL_setup.exe, follow the simple steps below the image to resolve.
Right-click the VIZYUL_setup.exe file and click Properties
In the VIZYUL_setup.exe Properties dialog box, click the Unblock button, click Apply and finally click the OK button. (see screen shot below)
After clicking OK, you can now double-click the VIZYUL_setup.exe file to install the software.
If you receive the following message while attempting to install VIZYUL_setup.exe, simply click Run to continue the installation.
After double-clicking the installation file, you’ll see the following screens. Click Next twice and you’ll be presented with a security dialog box requesting your approval to install the software. (see Windows 7 and 8 security dialog prompts).
Workbook Discovery is one of VIZYUL’s productivity features. This process automatically scans common folders for tableau workbooks on your computer. Once discovered, the Workbooks will appear under the Workbooks Folder and Local Computer section of the Workbook Selection pane on the left. VIZYUL automatically sorts Workbooks with the most recently accessed Workbooks at the top. If no Workbooks are found on the local computer the Workbooks Folder and Local Computer sections will be blank.
Opening VIZYUL will automatically query VIZYUL servers for the latest version of the software. You can choose to accept the new software installation by clicking the Update button. You also have the options to postpone or skip the upgrade.
Workbook Discovery is one of VIZYUL’s productivity features. The first time VIZYUL opens, it scans common folders for tableau workbooks. Once discovered, the Workbooks will appear under the Workbooks Folder and Local Computer sections of the Workbook Selection pane on the left. VIZYUL automatically sorts Workbooks with the most recently accessed Workbooks at the top. If no Workbooks are found on the local computer the Workbooks Folder and Local Computer sections will be blank.
After the initial opening, VIZYUL automatically syncs Workbooks with the latest version on the local computer. Yes, after updating your Workbooks with Tableau, VIZYUL will position those Workbooks at the top of the Workbooks Folder and Local Computer sections.
-
- Wizard Mode (default) – Guided navigation mode ideal for VIZYUL beginners. An easy three step process guides the user through selecting Workbooks to analyze, selecting the rules that will be run to analyze the Workbooks and finally reviewing the outcome of the analysis on the Analysis Overview screen.
- Expert Mode – Flexible, self-directed interface designed for VIZYUL power users. For example, the Tableau Administrator that wants to create a group of Custom Defined Rules that will be used later to analyze Workbooks.
Wizard Mode
Expert Mode
Drag-n-Drop – Drag Workbooks from any accessible folder on your computer right onto the VIZYUL logo.
Browser for Workbooks – Locate Workbooks to analyze by searching the drives and folders on your local computer.
Search for Workbooks – Search the Workbooks discovered during the workbook discovery process and workbooks on the local computer. Includes an auto-complete feature.
Workbooks Folder – Lists Workbooks discovered in the Workbook folder of the Tableau repository. The My Tableau Repository is located in your My Documents folder when Tableau Desktop Professional is installed on a computer.
Local Computer – Lists Workbooks discovered in My Documents and the current user’s Desktop folder.
Recent Workbooks – Contains a list of recently Analyzed Workbooks.
Peak Into Workbooks – When a Workbook is selected for analysis, the Workbook selection screen provides high-level insight into the number of data sources in the Workbook, the number of dashboards, the number of Worksheets, and the tableau version.
The built-in Best Practice and Performance Tuning rules are designed to unveil improvement opportunities for Tableau Workbooks. The Best Practice and Performance Tuning rules are not designed to communicate what’s wrong with Workbooks, but potential opportunities than can render improvements to the design and performance of your Workbooks.
A clear example of this is the Large Extracts built-in Performance Tuning rule. This rule fires for any Workbook with a data source that uses an extract with more than 500,000 rows of data. However, even though this rule fired, it doesn’t mean you should never have a Workbook where a data source has more than 500,000 rows of data. It means if performance is an issue for a Workbook with one or more large extracts, there may be an opportunity to improve performance by assessing this area.
Clicking a rule category opens the Rule toggle interface. This is where you can disable some or all rules that will be used to analyze Workbooks.
- Best Practice – Recommendations and warning severity rules designed to give insight into Workbook design opportunities.
- Performance Tuning – High, medium and low severity rules designed to give insight into Workbook performance opportunities.
- Custom Defined Rules – Define custom rules to analyze Workbooks
- Create Custom Rule screen – Interface to define custom rules.
- Disabling All Rules – If you don’t need Best Practice and/or Performance Tuning analysis for Workbooks, all rules can be disabled.
Home Screen – Groups rules into Rule Category and Severity sections/bars. Click a bar or an icon to look into which rules fired when analyzing the selected Workbooks.
- Analyzed Workbooks – Vertical tile view of all Workbooks selected for analysis. Clicking a Workbook will drill into the rules that fired for that Workbook.
- Breadcrumbs – Provides a means of navigating to different views available as part of the Workbook Analysis.
- Rule Category – Best Practice, Performance Tuning and Custom Defined Rules.
- Rule Severity
- Best Practice rules are grouped into recommendation and warning severity.
- Performance Tuning rules are grouped into high, medium and low severity. Severity indicates the prospective impact of performance improvements. For example, when a high severity rule fires, it’s likely that significant performance improvements are possible.
What’s a Session? – A VIZYUL package containing the Workbooks selected, the rules selected and the outcome of the analysis each time the application is used.
The Open Session window appears when one or more VIZYUL sessions are created.
Complete sessions have Workbook(s) selected for analysis, one or more rules selected and that have completed the analysis process.
Incomplete sessions haven’t completed the analysis process.
Recent Sessions – Lists both completed and incomplete sessions. Double-click a session in the recent session or the tile view area to open a session.
Sessions Tile View – Contains only a list of complete sessions. Double-click a session in the recent session or the tile view area to open a session.
Rules Engine
-
Rationale
The VIZYUL Live Connections or Extracts best practice rule fires for each Workbook that has one or more live data connections.
Insight
The decision on whether to use live connections or extracts in the life cycle of authoring tableau dashboards can have a direct impact on your tableau experience and ultimately the experience of your intended audience. At VIZYUL, tableau dashboard design happens in three phases; data discovery, cognitive clarity and dashboard refinement. Each phase can have a wide variety of data needs. However, once we’re happy with the final version of our dashboard, that’s when we look to trim the optimize live data connections and extracts for optimal performance. Here are a few things to consider.
Data Discovery
In this phase, you’re not quite sure what data you’ll need for the final dashboard, so you explore various data sets with tableau. In this phase, we recommend using the tableau Data Engine to create tableau Data Extracts. Here’s why. Using tableau data extracts reduces the the workload on corporate information data systems. Tableau Data Extracts are phenomenally fast so performance is not a hindrance to your creativity. Finally, tableau Data Extracts expose all of the wonderful features available in tableau desktop. Some Live Data Connections limit the features in tableau desktop; like a live connection to a Microsoft Access database.
Cognitive Clarity
Now that you’ve seen and understood your data visually, you’re ready to fine tune the story being told by the data by visualizing it. VIZYUL™ tableau authors use this phase to optimize the data sources including only the dimensions, measures and calculated fields needed to tell the visual story. This is the ideal time in the evolution of your dashboard to begin to make decisions about the final type of data connections your dashboard will use. Data connection types can have a dramatic impact on the overall performance of your dashboard. Here are a few things to consider during this process.
- If the final Dashboard uses a high performance analytic data source? – Consider using a Live Data Connection
- If the final dashboard uses a data source with tons of data that you really don’t need for your dashboard? – Consider using a tableau Data Extract
- Does your dashboard need real-time access to the data being visualized? – Consider using a Live Data Connection with custom query to pull only the data you need
- Does your dashboard only require a small portion of the data available in the data source? – Consider using a tableau Data Extract
- Have your tableau administrators setup data sources on a tableau server? – Consider using a live tableau server data connection
In our experience, the final decision on the appropriate type of data connection involves the data consumer (you) and the data provider (usually IT or the data management group). So before you make the final decision on the type of data connection for your dashboard, it’s always a good idea to partner with IT to make sure your dashboard performs well over time.
Dashboard Refinement
Finally, it’s time to show your dashboard to the world! YEAH! The right data connection types used by your dashboard can mean the difference between a tableau experience people love and want to use or one that creates angst.
Action
- Consider carefully whether to use live connections or extracts in your dashboards.
- Consider the resources below
Additional Resources
Live or Extract Decision Making
- http://www.tableau.com/learn/whitepapers/memory-or-live-data
- http://www.tableau.com/about/blog/2014/7/understanding-tableau-data-extracts-part1
- http://www.tableau.com/about/blog/2014/7/why-use-tableau-data-extracts-32187
- http://www.tableau.com/tableau-data-extracts-part3
- http://mindmajix.com/use-direct-connection-data-extract-tableau/
- https://slalomdotcom.wordpress.com/2014/03/26/how-to-improve-the-performance-of-your-tableau-dashboards/
Getting Familiar with Tableau Data Connections
- http://kb.tableau.com/articles/knowledgebase/export-data-connection
- http://community.tableau.com/message/286629
Rationale
The VIZYUL™ Reduced Tableau Features best practice rule fires when a dashboard uses a live data connection to a Microsoft Access database.
Insight
So, there’s this thing called a data access library. The data access library defines how your tableau dashboard accesses the data it uses for the visualizations you create. Unfortunately data access libraries offer varying degrees of capabilities when used for tableau data connections.
Microsoft Access, Excel and Text file data connections all use the Microsoft Jet database engine data access library. As of the tableau 8.2 release, using a live connection to a Microsoft Access database disables the following features in tableau.
- Tableau 8.1 or prior – Count Distinct (unique count of dimensions or measures), Media and Percentiles are disabled
- Tableau 8.2 – Count Distinct (unique count of dimensions or measures)
- Tableau 9 – Count Distinct (unique count of dimensions or measures)
Action
If the need arises to use data from Microsoft Access databases, consider creating a tableau data extract by importing the data from Microsoft Access into your Tableau workbook.
Additional Resources
- http://kb.tableau.com/articles/knowledgebase/accessing-countd
- http://drawingwithnumbers.artisart.org/i-have-wee-data-microsoft-access-and-tableau/
- http://community.tableau.com/thread/146010
Microsoft Access Data Connections
- http://onlinehelp.tableau.com/current/pro/online/windows/en-us/examples_access.html
Rationale
The VIZYUL™ Automatic Dashboard Sizing best practice rule fires for each dashboard that uses the automatic sizing setting.
Insight
The Automatic Dashboard size setting allows you to create a Dashboard that fills the screen of the dashboard viewer.
Consider this, dashboards that use automatic dashboards sizing adapt to the amount of screen space available on the device used to view the dashboard. Basically this means unless everyone viewing your dashboard has the exact same resolution, it’s quite likely that your dashboard will re-size itself in an unattractive and possibly incomprehensible way.
Action
VIZYUL™ Tableau authors recommend considering the type of devices your audience will use while viewing your dashboard. Once you know the range of these devices and their screen resolutions, you’re armed to make a quality decision on the appropriate size of your dashboard.
Additional Resources
- http://kb.tableau.com/articles/knowledgebase/fixed-size-dashboard
- http://kb.tableau.com/articles/knowledgebase/best-practices-designing-vizes-and-dashboards
Rationale
The VIZYUL™ Consider Highlight Actions best practice rule fires when a dashboard doesn’t use highlight actions.
Insight
Consider this, dashboards that empower viewers to ask additional questions about the information presented in the dashboard is the easiest way to increase adoption. Tableau highlight actions are a fast and simple way to add this level of interactivity to your dashboards.
With highlight actions enabled on your dashboards you empower users to narrow their focus on aspects of your dashboard important to them.
Action
- Consider using highlight actions to add additional interactivity to your dashboards.
- Consider the resources below
Additional Resources
Understanding Highlight Actions
- http://downloads.tableau.com/quickstart/feature-guides/actions_highlight.pdf
Great Examples of Highlight Actions
- http://paintbynumbersblog.blogspot.com/2014/10/a-rough-guide-to-tableau-dashboard.html
Rationale
The VIZYUL™ Consider Filter Actions best practice rule fires when a dashboard doesn’t use filter actions.
Insight
Consider this, dashboards that empower viewers to ask additional questions about the information presented in the dashboard, by drilling down to deeper levels of insight, is one of the easiest ways to increase adoption. Tableau filter actions are a fast and simple way to add this level of interactivity to your dashboards.
With filter actions enabled on your dashboards you empower users to filter the data at the level important to them.
Action
- Consider using filter actions to add drill down or deep dive interactivity to your dashboards.
- Consider the resources below
Additional Resources
Understanding Filter Actions
- http://kb.tableau.com/articles/knowledgebase/creating-filter-actions-dashboards
- http://paintbynumbersblog.blogspot.com/2014/10/a-rough-guide-to-tableau-dashboard.html
Great Examples of Filter Actions
- Creative Filter Actions Using Images – http://www.evolytics.com/blog/tableau-201-create-icon-based-navigation-filters/
Rationale
The VIZYUL™ Consider Annotations best practice rule fires when annotations have not been used on a dashboard.
Insight
Consider this, annotations are an excellent way to draw viewers attention to areas of importance on your dashboards. At VIZYUL™ we ask ourselves a simple question to determine whether a dashboard is effectively communicating the intended subject matter. “How many seconds does it take a viewer to have that a ha! moment when viewing a dashboard”? If it takes more than 3 to 5 seconds, we know there’s room to improve the cognitive clarity of the dashboard.
Annotations are an easy and excellent way to avoid authoring difficult to understand dashboards.
Action
- Consider using annotations on dashboards with unexplained outliers, or simply to improve the clarity.
- Consider the resources below
Additional Resources
Understanding Annotations
- http://onlinehelp.tableau.com/current/pro/online/en-us/annotations_annotations_add.html
Great Examples of Annotations
- Annotate with values not on the worksheet – http://drawingwithnumbers.artisart.org/annotating-a-view-with-a-total/
- http://blog.visual.ly/wp-content/uploads/2012/04/tahrirDiagram_800.jpeg
- This example isn’t easy in tableau but a great example none the less – http://www.nytimes.com/interactive/2011/05/03/us/20110503-osama-response.html
- http://projects.nytimes.com/guantanamo
Rationale
The VIZYUL™ Unused Measures best practice rule fires when a measure in a data source isn’t used on any worksheet.
Insight
Unused measures are frequent occurrence here at VIZYUL™. Many times it’s necessary to include fields that ultimately won’t be used and create calculated fields while performing data discovery activities. However, if we’re in the process of finalizing the dashboard, unused measures can negatively impact the overall performance of the dashboard.
It’s important to note that tableau includes a “Hide Unused Fields” feature. While hiding unused fields are a way to improve performance, experts advise removing fields you know aren’t used when authoring dashboards.
Action
- Consider using removing all unused measures from each data source in your workbook. NOTE: Tableau will warn or prevent you from accidentally removing measures that are used in a workbook.
- Consider the resources below
Additional Resources
- How To – http://onlinehelp.tableau.com/current/pro/online/mac/en-us/datafields_dwfeatures_hide.html
- Tips – http://kb.tableau.com/articles/knowledgebase/tips-working-with-extracts
- Tips – http://kb.tableau.com/articles/knowledgebase/controlling-displayed-fields
Rationale
The VIZYUL™ Unused Dimensions best practice rule fires when a dimension in a data source isn’t used on any worksheet.
Insight
Unused dimensions are frequent occurrence here at VIZYUL™. Many times it’s necessary to include fields that ultimately won’t be used and create calculated fields while performing data discovery activities. However, if we’re in the process of finalizing the dashboard, unused dimensions can negatively impact the overall performance of the dashboard.
It’s important to note that tableau includes a “Hide Unused Fields” feature. While hiding unused fields are a way to improve performance, experts advise removing fields you know aren’t used when authoring dashboards.
Action
- Consider using removing all unused dimensions from each data source in your workbook. NOTE: Tableau will warn or prevent you from accidentally removing dimensions that are used in a workbook.
- Consider the resources below
Additional Resources
- How To – http://onlinehelp.tableau.com/current/pro/online/mac/en-us/datafields_dwfeatures_hide.html
- Tips – http://kb.tableau.com/articles/knowledgebase/tips-working-with-extracts
- Tips – http://kb.tableau.com/articles/knowledgebase/controlling-displayed-fields
Rationale
The VIZYUL™ Dashboard Complexity best practice rule fires when a dashboard has more than four worksheets on it.
Insight
Consider this, dashboards with many worksheets allow authors to include many dimensions and measures on the same dashboard. Obviously, the benefit of this design choice is centralized content. At first glance this may not seem like a problematic design choice, however, here are a few things to consider before settling on this approach.
Performance Implications
- Each worksheet and quick filter added to a dashboard triggers a tableau query against the data source equal to the number of objects on the dashboard.
- Each time tableau has to query a data source, doesn’t matter whether its a live or extract connection, there is a load time cost.
- The more worksheets on a dashboard, the longer it will take tableau to render the view (deliver the visualization to your viewers). This is determined by the number of queries tableau has to execute and the time cost for each query.
- Tableau 8.3 or prior executes each query synchronously, which means each query must finish before the next run is executed.
How Tableau Improves Performance
- Caching – Basically this means storing the aggregated data you need so additional queries aren’t necessary. This only applies to the dimensions and measures used on specific worksheets to optimize performance.
- Parallel Queries – In tableau version 9 tableau performs queries in parallel. Basically this means non-cached queries are executed at the same time for different data sources at the same time, cutting down on the time it takes to render the view.
Action
- Consider reducing the number of worksheets on a dashboard.
- Consider the order in which worksheets and quick filters are added to the dashboard. The order has a direct impact on tableau’s ability to leverage caching to improve the performance of your dashboard. See the first resource below for more info.
Additional Resources
- http://tableaulove.tumblr.com/post/98949091175/decrease-tableau-dashboard-render-time-with
- http://kb.tableau.com/articles/knowledgebase/database-query-performance
- Video (requires free tableau login to view) – http://www.tableau.com/learn/tutorials/on-demand/tableau-server-authoring-performance
- http://kb.tableau.com/articles/knowledgebase/optimizing-tableau-server-performance
- http://www.tableau.com/about/blog/2015/1/90-preview-query-performance-improvements-36406
Rationale
The VIZYUL™ Extract Refresh best practice rule fires when an extract hasn’t been refreshed in five or more days.
Insight
Consider this, when you extract data for use in a tableau dashboard, you are effectively taking a snapshot of the data. By snapshot we mean a static copy of the data that is frozen in time. The only exception to this is if you’ve published your dashboard to a tableau server and elected to set an automatic extract refresh schedule. Otherwise, the extract is a snapshot of data frozen at the time the extract was taken.
Seeing this rule fire doesn’t necessarily mean your dashboard isn’t adhering the best practice techniques. This rule is designed to be used as a “friendly reminder”. Static extracts are perfect for the data discovery phase of dashboard design. However, when you dashboard is ready for prime time, you may want to leverage this rule as a reminder to ensure regular updates are in place for data extracts.
Action
- Before debuting a dashboard, consider refreshing data extracts on a regular schedule that makes sense for your purposes.
- Consider the additional resources below.
Additional Resources
- http://kb.tableau.com/articles/knowledgebase/refresh-extract
- http://drawingwithnumbers.artisart.org/o-extract-where-art-thou/
- Tableau Server v9 – http://onlinehelp.tableau.com/current/server/en-us/help.htm#qs_refresh_extracts.html
Rationale
The VIZYUL™ Unused Data Sources best practice rule fires when a workbook contains unused data sources.
Insight
Consider this, during the data discovery phase of designing a dashboard, often times data sources are added to a tableau workbook that don’t end up making the final cut. Which means there are unused data sources in your workbook.
Unused data sources usually don’t negatively impact the time it takes tableau to render your dashboards, well, because it’s not used. However, there are other areas where unused extracts can have an unfavorable impact of a tableau workbook.
Consider the case where a tableau user has finalized a dashboard and wants to share it with the team. The author plans to save the workbook as a packaged tableau workbook. If the unused data sources are extracts, each one increases the size of the packaged tableau workbook. In this case, removing unused data sources can have a positive impact on the size of the overall workbook.
Action
- Consider removing unused data sources from your workbook prior to publishing or giving access to your viewers.
- Consider the additional resources below
Additional Resources
- https://slalomdotcom.wordpress.com/2014/03/26/how-to-improve-the-performance-of-your-tableau-dashboards/
- 3 Part Series on Understanding Tableau Data Extracts
- http://www.tableau.com/about/blog/2014/7/understanding-tableau-data-extracts-part1
- http://www.tableau.com/about/blog/2014/7/why-use-tableau-data-extracts-32187
- http://www.tableau.com/tableau-data-extracts-part3
Rationale
The VIZYUL™ Consider Formatted Tooltips best practice rule fires when a dashboard has no formatted tooltips.
Insight
Consider this, formatted tooltips can play a very important role of designing a compelling, engagement and informative dashboard. Here are just a few ways formatted tooltips can be used to improve your dashboards.
- Conditional formatted messages/indicators i.e. green text for above threshold and red for below threshold
- Enrich dashboards by including relevant and contextual data in formatted tooltips
- Notes on the data used on the worksheet
- Stating sources of data
- A means of inviting viewers to interact with the dashboard
- A great way to grab the attention of the viewer
Consider the following comparison. Which one is more likely to grab your attention?
Action
- Consider using formatted tooltips to grab the attention of users or to invite interaction.
- Consider the additional resources below
Additional Resources
- Video – Conditionally Formatting Tooltips – http://kb.tableau.com/articles/knowledgebase/format-tooltips
- https://www.dataz.io/display/Public/2014/04/28/Tableau+Pro+Tip%3A+Adding+Icons+to+Tables+and+Tooltips
- http://www.theinformationlab.co.uk/2014/05/08/tooltips-tableau-dashboards-user-experience-perspective/
Rationale
The VIZYUL™ Pie Charts with More Than Four Slices best practice rule is pretty straight forward.
Insight
Consider this, the human eye can perceive many different colors. However, distinguishing many different shades of color in objects in the same object becomes exponentially more difficult. Not sure this is true? Tell me how many colors are in the pie chart below. LOL..okay, that’s an unfair example, but I think you get the point.
Tableau built-in color palettes don’t extend beyond 20 different colors because it’s challenging for us to distinguish between them. Making it nearly impossible for us to accurately assess what an object with that many colors wants us to glean. Because of this reality, we recommend considering how may slices and colors are used in pie charts.
Action
- Consider restricting pie charts to no more than 6 slices and different colors.
- Consider the additional resources below
Additional Resources
- http://www.theinformationlab.co.uk/2015/01/06/show-pie-charts/
- http://tblsft.com/public/community/best-practices/pie-chart
- https://eagereyes.org/techniques/pie-charts
Rationale
The VIZYUL™ Published Workbook with Local Data Sources best practice rule fires when a workbook has been published to a Tableau server but references data sources on the computer where the workbook was designed.
Insight
Consider this, publishing a tableau workbook to a tableau server, when one or more of the data connections point to data sources on your local computer, will prevent tableau server from updating the data source automatically when new data becomes available.
This is a rather simple concept to understand. Let’s say you have an Excel spreadsheet in your My Documents folder that you want to use in tableau to design a dashboard. So you point tableau at your Excel file and import the data; creating a tableau data extract. Great so far. However, it’s important to note that even though you imported the data into tableau, tableau keeps an internal record of the original location of the Excel file you used. You finish your dashboard and decide to publish the workbook to tableau server for the rest of your team to view. Everything seems to work fine until a college asks you to schedule a refresh of the data every Sunday morning on tableau server.
You successfully setup the automated refresh schedule only to find on Monday morning that the refresh crashed Sunday morning. The refresh failed because tableau server has no way to access the original Excel file located in your My Documents folder. The reason your tableau workbook recorded the original location of this file is to tell tableau server where to look in order to refresh the data.
Here’s a checklist that will help determine which type of data connection supports automated data refreshes on the tableau server.
- Data Connection Types
- Live Database Connection (SQL Server, Oracle, TeraData, MySQL, PostgreSQL, etc…)
- NO – Live database connections do not extract data from the origin database. The data used to render the dashboard is also fresh.
- Live File Connection (Excel, Access, Text File, CSV, etc…)
- NO – Although tableau desktop or tableau server will prevent you from publishing the workbook, by creating data extracts for all data connections, refreshing the data will fail on tableau server, if the exact folder location of the original file is not replicated on the tableau server.
- Tableau Data Extract from Database (SQL Server, Oracle, TeraData, MySQL, PostgreSQL, etc…)
- YES – Automated data refreshes are supported (assuming the tableau server has the necessary permissions to access the database server directly)
- Tableau Data Extract from Local File (Excel, Access, Text File, CSV, etc…)
- NO – Refreshing the data will fail on tableau server, if the exact folder location of the original file is not replicated on the tableau server.
- Live Tableau Server Data Source
- YES – Assuming the tableau server data source has been configured to access the original data source and has the necessary permissions to access it for data refreshes.
- Live Database Connection (SQL Server, Oracle, TeraData, MySQL, PostgreSQL, etc…)
Action
- Consider publishing all data sources that require automated refreshes to tableau server.
- Consider the additional resources below for information on how to prepare data extracts for automated refreshes.
Additional Resources
- http://onlinehelp.tableau.com/current/pro/online/windows/en-us/extracting_push.html
- http://onlinehelp.tableau.com/current/pro/online/mac/en-us/extracting_refresh.html
- 3 Part Series on Understanding Tableau Data Extracts
- http://www.tableau.com/about/blog/2014/7/understanding-tableau-data-extracts-part1
- http://www.tableau.com/about/blog/2014/7/why-use-tableau-data-extracts-32187
- http://www.tableau.com/tableau-data-extracts-part3
-
Rationale
The VIZYUL™ Large Extract performance tuning rule fires tableau data extracts with 500,000 or more rows of data.
Insight
Consider this, during the data discovery phase of designing a dashboard, often times data sources are added to a tableau workbook that don’t end up making the final cut. Which means there are unused data sources in your workbook.
Unused data sources usually don’t negatively impact the time it takes tableau to render your dashboards, well, because it’s not used. However, there are other areas where unused extracts can have an unfavorable impact of a tableau workbook.
Consider the case where a tableau user has finalized a dashboard and wants to share it with the team. The author plans to save the workbook as a packaged tableau workbook. If the unused data sources are extracts, each one increases the size of the packaged tableau workbook. In this case, removing unused data sources can have a positive impact on the size of the overall workbook.
Action
- Consider removing unused data sources from your workbook prior to publishing or giving access to your viewers.
- Consider the additional resources below
Additional Resources
- How to publish ONLY the metadata for large tableau data extract (if you have a billion row extract you need to get to the tableau server without having your poor laptop process that much data, this post is FOR YOU!)
- Method 1 – http://tableaulove.tumblr.com/post/18945358848/how-to-publish-an-unpopulated-tableau-extract
- Method 2 – http://www.tableau.com/about/blog/2013/9/easy-empty-local-extracts-25152
- http://kb.tableau.com/articles/knowledgebase/optimizing-incremental-refreshes
Rationale
The VIZYUL™ Calculated Fields performance tuning rule fires for any data connection with 7 or more calculated fields.
Insight
Consider this, the number of calculated fields in a data connection can have a direct impact on the overall performance of the workbook. Specifically the calculated fields used on worksheets and those that are calculated at the level of detail.
In order to better understand how calculated impact performance, we’ll walk thru a simple scenario. My original data source contains three fields, sales, region and items (the number of items sold for that region). The entire data set has 10 rows of data, keeping it simple. If I create a calculated field that evaluates the value of each region and then either divides sales by items or multiplies sales time items, I’ve created what’s commonly referred to as a LOD (level of detail) calculation. Basically LOD calculation means a field where the calculation must be performed for every row in the data set.
As you can imagine, if I drop my new calculated field on a worksheet, it’ going to take tableau longer to aggregated my new field than it would to simply aggregate sales. For this reason, this rule simply alerts the VIZYUL™ user to an opportunity to optimize the data connection for optimal performance.
Here are a few techniques you can use to optimize calculated fields, based on your specific environment.
- I have access to a tableau server to publish my data sources
- Create Calculations in the Original Data Source – Offloading the processing of calculated fields onto the origin data source is our preferred approach. However, we also understand that this method isn’t available to everyone. For Excel users this means adding an additional column to your spreadsheet that contains the calculation prior to connecting to the spreadsheet with tableau desktop. You’ll be surprised the performance gain you’ll experience taking this simple step. For those accessing data on a database server, this usually means a custom sql query that calculates the desired fields.
- Optimize Calculations – We recommend, as much as possible, creating calculations that use data that tableau has already aggregated. An example would be the difference between SUM([sales])/SUM([items]) versus sales/items. The former causes tableau to aggregate all sales and items before performing the division operation. The latter is performed for each row of data included in the worksheet.
- Use Tableau Desktop’s Optimize Feature (only available for tableau data extracts) – Performing this on a data source causes tableau desktop to generate metadata for the calculated fields you’ve created. For measures tableau desktop internally stores things like the minimum and maximum values. For dimensions tableau desktop internally stores the unique set of values for the dimension. Storing this metadata increases performance by providing information that tableau uses to query the data. See the screen shot for the location of this feature.
- Publish Calculated Fields to Tableau Server – Let’s say you’ve created 50 calculated fields; all of which you need for your dashboards. If you don’t publish the data source to tableau server, each time you view one of your dashboards, tableau desktop has to calculate each of the calculated fields on the worksheets included on the dashboard. However, if you publish the data source to tableau server, tableau is smart enough to create actual fields from your 50 calculated fields. This has the same or better effect as if you generated the calculated fields prior to bringing the data into tableau desktop. This will always have an immediate positive impact on the overall performance of your dashboards.
- PLEASE NOTE: There are a few scenarios where, even if you publish a data source to tableau server, the calculated field will NOT be turned into an actual field yielding optimal performance. As of this writing this includes ANY calculated field that includes a parameter. This is because it difficult to determine the actual result of a calculation based on a parameter since the parameter can change at any time. Calculated fields that use NOW(), TODAY(), USER() or USERDOMAIN() also cannot be materialized (turn into an actual field). The functions mentioned are specific to a given computer so it’s difficult to determine the actual value of the calculated field.
- I DO NOT have access to a tableau server to publish my data sources
- See 1.1 above
- See 1.2 above
- See 1.3 above
Action
- Consider the recommended steps above to optimize your data source calculated fields.
- Consider the additional resources below
Additional Resources
- http://community.tableau.com/message/221542
- http://onlinehelp.tableau.com/current/pro/online/windows/en-us/help.htm#extracting_optimize.html?Highlight=optimizing
- https://boraberan.wordpress.com/2015/01/30/whats-new-in-tableau-9-0-part-2-level-of-detail-expressions/
Rationale
The VIZYUL™ Large Extract Data Blends performance tuning rule fires for tableau data extracts with 2,000,000 or more rows of data that use data blends.
Insight
Consider this, data blends are a powerful feature of tableau desktop. This feature allows you to design dashboards with various data sources right inside tableau desktop. Often, data blends are the most convenient means of adding context and richness to data visualizations.
That said, data blends on large tableau data extracts can degrade performance relatively quickly. This is because data blends force tableau to evaluate matching values between the data sources for each row included in the worksheet. So if the main (primary – blue check mark) data source used by the worksheet has 20 million rows of data and the blended (secondary – orange check mark) data source has 100 rows, that means tableau desktop has to make 2 billion value comparisons. Obviously, if a data blend can be avoided in this scenario, this would have an immediately positive impact on performance.
Here are a few alternatives to data blends, when large extracts are involved.
- When possible, use data joins at the origin data sources. Instead of doing a data blend in tableau desktop, use SQL to join multiple tables prior to bringing the data into tableau desktop.
- Use queries in Microsoft Access to join multiple tables prior to bringing data into tableau desktop.
Action
- Consider using data joins outside of tableau desktop or data joins within tableau when using tableau data extracts with 2,000,000 or more rows of data.
- Consider the additional resources below
Additional Resources
- http://kb.tableau.com/articles/knowledgebase/join-vs-relationship
- http://onlinehelp.tableau.com/v6.1/public/online/en-us/i1003860.html
- Michael Sandberg’s Series on Data Blending
- http://datavizblog.com/2014/04/04/an-introduction-to-data-blending-part-1/
- http://datavizblog.com/2014/04/06/an-introduction-to-data-blending-part-2-hans-rosling-gapminder-and-data-blending/
- http://datavizblog.com/2014/04/08/an-introduction-to-data-blending-part-3-benefits-of-blending-data/
- http://datavizblog.com/2014/04/13/an-introduction-to-data-blending-part-4-data-blending-design-principles/
- http://datavizblog.com/2014/04/28/an-introduction-to-data-blending-part-5-tableaus-data-blending-architecture/
Rationale
The VIZYUL™ Complex Calculated Fields performance tuning rule fires for any calculated field with 400 or more characters.
Insight
Consider this, data blends are a powerful feature of tableau desktop. This feature allows you to design dashboards with various data sources right inside tableau desktop. Often, data blends are the most convenient means of adding context and richness to data visualizations.
That said, data blends on large tableau data extracts can degrade performance relatively quickly. This is because data blends force tableau to evaluate matching values between the data sources for each row included in the worksheet. So if the main (primary – blue check mark) data source used by the worksheet has 20 million rows of data and the blended (secondary – orange check mark) data source has 100 rows, that means tableau desktop has to make 2 billion value comparisons. Obviously, if a data blend can be avoided in this scenario, this would have an immediately positive impact on performance.
Here are a few alternatives to data blends, when large extracts are involved.
- When possible, use data joins at the origin data sources. Instead of doing a data blend in tableau desktop, use SQL to join multiple tables prior to bringing the data into tableau desktop.
- Use queries in Microsoft Access to join multiple tables prior to bringing data into tableau desktop.
Action
- Consider using data joins outside of tableau desktop or data joins within tableau when using tableau data extracts with 2,000,000 or more rows of data.
- Consider the additional resources below
Additional Resources
- Outstanding Tips and Tricks for Calculations by Alan Eldridge from the Drawing with Numbers Blog – http://drawingwithnumbers.artisart.org/wiki/tableau/tricks-miscellaneous-techniques/
-
Rule Name (required)
Rule name is a required field and the ideal place to give a meaningful name to a custom analysis rule ...
Read MoreRule Description (required)
Rule description is a required field and ideal for specifying the conditions required for the rule to fire ...
Read MoreRule Category
Rule category is an optional field designed to group custom analysis rules into logical divisions. When reviewing the custom analysis ...
Read MoreRule Severity
Rule severity allows you to categorize rules into groupings important to your business ...
Read MoreRule Targets
Rule targets represent design attributes of a tableau workbook. Basically this is the attribute of your Tableau dashboard your custom ...
Read More