Tableau Software has finally released the new version of its benchmark setting data visualization and business intelligence platform, Tableau 9.0 today on April 7th 2015. The new version comes with an entirely new experience for Tableau Server, significant performance improvements, new data preparation features, drag-and-drop analytics, updates to mapping and much more.
Tableau 8 has been a phenomenal success[1] for data visualization and business intelligence platform provider Tableau Software after the release of Tableau 8.3, everybody was waiting for the new version : Tableau 9. Tableau 9 release date is postponed a few times. In Tableau Conference 2013, the expectation was that Tableau 8.1 would be released in fall 2013, Tableau 8.2 in winter 2014 and Tableau 9.0 later in 2014. Tableau 9.0 release date is later pushed to Q1 2015 and in March one more time to April 2015.
But this is a release worth waiting. There are many significant improvements in Tableau 9.0 over Tableau 8.3 which makes already powerful Tableau Software more competitive.
Tableau User Experience Improvements
Ease of use and flawless data analysis process has always been in the core of Tableau and the company is wisely investing more in this core strength with exiting new features. One of the most significant new feature in the front-end is ad-hoc calculations. You can now write expressions directly on the rows and columns shelves or mark cards with a simple double click. You can later drag these ad-hoc calculations to Data Window to make them reusable calculations.
A new Analytics Pane in Tableau 9 makes common techniques such as reference lines, bands, totals, trend lines and forecasts readily available by a simple drag-and-drop. Another feature related to Tableau's analytics capabilities is Instant Analytics which provides an interactive experience for comparing summary information about subset of marks to all marks in the view.
New level of details calculation is another exiting new feature which allows you to create arbitrary aggregates in charts without complex and cumbersome expressions. For example, with this new feature, you can easily calculate average sale per customer in each time period by a simple, nested Level of Details Calculation (LOD). LOD syntax also allows you to create fixed values not effected by level of detail in the charts.
Data Connection and Preparation Improvements
Tableau 9.0 now supports Kerberos authentication for Microsoft SQL Server, SQL Server Analysis Service and Cloudera Impala. adds data extraction API for Mac OS and offers new and improved data connectors such as SPSS, SAS and R.
Recently, almost every Tableau release adds more data preparation functionality. Tableau 9.0 is not an exception. Tableau now can detect the range of data in MS Excel files, unpivot tables (converts a cross-tab table to straight table), split data in multiple fields.
Tableau 9 Performance Improvements
Tableau data engine, the in-memory analytics database of Tableau platform, was introduced in Tableau 6 and went through performance improvements since than. But Tableau 9 seems the be the biggest jump. In the presentation, a bar-chart aggregated from 173 million rows of data took 0.7 seconds in Tableau 9 compared to 7 seconds performance of Tableau 8.3 (10x faster)!
As this individual query performance is not enough. Tableau 9 data engine also introduces parallel queries concept. If you have a Tableau dashboard with 3 charts and you do something to trigger a query (i.e. filtering dashboard), each chart queries the database one after another in sequence. In the presentation, a 3 chart dashboard took 9.5 seconds to load. In Tableau 9 it takes 1.1 seconds. Not only individual queries are faster, they also run parallel (provided that underlying database supports parallel queries).
And then comes query fusion. Even if you run parallel queries, why to run a the same query twice if two charts in the view has similar queries which can be run once. Tableau Query Fusion looks at the dashboards and finds way to simply the queries into simple queries.
Suppose that you have two sheets in a dashboard : Sum of Rides per Hour and Average Tip per Hour. In Tableau 8.3, these two will fire 2 individual queries (in sequence) to database. But in Tableau 9, because these two has the same level of details, they will be fused into a single query and run fast. A sample run for 173 million rows takes 3.7 seconds in Tableau 8.3 but only 1.7 seconds in Tableau 9 thanks to query fusion.
So Tableau 9 offers faster queries, parallel queries and thanks to Query Fusion, less queries. And thanks to new External Query Catching, it also offers no queries. Tableau 9 will cash queries in server and desktop and run no queries to database if there is no chance there. In an example run, a 3 sheet dashboard running 9.7 seconds in Tableau 8.3 took only 0.2 seconds to run. 50x improvements.
One of the most desired feature for Tableau 9 is visual ETL (Extraction Transformation and Load) functionality for data quality, validation and cleaning (and data modeling if possible). Tableau 8.2's "visual data window" seems to be the first step to that direction but we will wait and see if this would be a new feature. Although there are very powerful visual ETL tools which can work seamlessly with Tableau such as Alteryx and Clover ETL, it would be nice to see some more features in this domain with Tableau 9.
On the predictive analytics domain, we expect more functionality in Tableau such as more predictive models and control in newly introduced forecasting feature and better integration with R (current Tableau version can do 4 functions passing parameters to R Server).
We will update this post as more information is revealed about Tableau 9.
[1] - Tableau's Building the 'Google for Data'
[2] - Tableau Gives Up Gains: Estimates Rise, Targets Decline, on Upbeat Q1, Forecast
[3] - Tableau Conference 2014 – A Field Report
[4] - Data Visualization with Tableau
Tableau 8 has been a phenomenal success[1] for data visualization and business intelligence platform provider Tableau Software after the release of Tableau 8.3, everybody was waiting for the new version : Tableau 9. Tableau 9 release date is postponed a few times. In Tableau Conference 2013, the expectation was that Tableau 8.1 would be released in fall 2013, Tableau 8.2 in winter 2014 and Tableau 9.0 later in 2014. Tableau 9.0 release date is later pushed to Q1 2015 and in March one more time to April 2015.
But this is a release worth waiting. There are many significant improvements in Tableau 9.0 over Tableau 8.3 which makes already powerful Tableau Software more competitive.
Tableau User Experience Improvements
Ease of use and flawless data analysis process has always been in the core of Tableau and the company is wisely investing more in this core strength with exiting new features. One of the most significant new feature in the front-end is ad-hoc calculations. You can now write expressions directly on the rows and columns shelves or mark cards with a simple double click. You can later drag these ad-hoc calculations to Data Window to make them reusable calculations.
 |
| Tableau 9 Ad-hoc calculations feature |
A new Analytics Pane in Tableau 9 makes common techniques such as reference lines, bands, totals, trend lines and forecasts readily available by a simple drag-and-drop. Another feature related to Tableau's analytics capabilities is Instant Analytics which provides an interactive experience for comparing summary information about subset of marks to all marks in the view.
New level of details calculation is another exiting new feature which allows you to create arbitrary aggregates in charts without complex and cumbersome expressions. For example, with this new feature, you can easily calculate average sale per customer in each time period by a simple, nested Level of Details Calculation (LOD). LOD syntax also allows you to create fixed values not effected by level of detail in the charts.
Data Connection and Preparation Improvements
Tableau 9.0 now supports Kerberos authentication for Microsoft SQL Server, SQL Server Analysis Service and Cloudera Impala. adds data extraction API for Mac OS and offers new and improved data connectors such as SPSS, SAS and R.
Recently, almost every Tableau release adds more data preparation functionality. Tableau 9.0 is not an exception. Tableau now can detect the range of data in MS Excel files, unpivot tables (converts a cross-tab table to straight table), split data in multiple fields.
Tableau 9 Performance Improvements
Tableau data engine, the in-memory analytics database of Tableau platform, was introduced in Tableau 6 and went through performance improvements since than. But Tableau 9 seems the be the biggest jump. In the presentation, a bar-chart aggregated from 173 million rows of data took 0.7 seconds in Tableau 9 compared to 7 seconds performance of Tableau 8.3 (10x faster)!
As this individual query performance is not enough. Tableau 9 data engine also introduces parallel queries concept. If you have a Tableau dashboard with 3 charts and you do something to trigger a query (i.e. filtering dashboard), each chart queries the database one after another in sequence. In the presentation, a 3 chart dashboard took 9.5 seconds to load. In Tableau 9 it takes 1.1 seconds. Not only individual queries are faster, they also run parallel (provided that underlying database supports parallel queries).
And then comes query fusion. Even if you run parallel queries, why to run a the same query twice if two charts in the view has similar queries which can be run once. Tableau Query Fusion looks at the dashboards and finds way to simply the queries into simple queries.
Suppose that you have two sheets in a dashboard : Sum of Rides per Hour and Average Tip per Hour. In Tableau 8.3, these two will fire 2 individual queries (in sequence) to database. But in Tableau 9, because these two has the same level of details, they will be fused into a single query and run fast. A sample run for 173 million rows takes 3.7 seconds in Tableau 8.3 but only 1.7 seconds in Tableau 9 thanks to query fusion.
So Tableau 9 offers faster queries, parallel queries and thanks to Query Fusion, less queries. And thanks to new External Query Catching, it also offers no queries. Tableau 9 will cash queries in server and desktop and run no queries to database if there is no chance there. In an example run, a 3 sheet dashboard running 9.7 seconds in Tableau 8.3 took only 0.2 seconds to run. 50x improvements.
One of the most desired feature for Tableau 9 is visual ETL (Extraction Transformation and Load) functionality for data quality, validation and cleaning (and data modeling if possible). Tableau 8.2's "visual data window" seems to be the first step to that direction but we will wait and see if this would be a new feature. Although there are very powerful visual ETL tools which can work seamlessly with Tableau such as Alteryx and Clover ETL, it would be nice to see some more features in this domain with Tableau 9.
On the predictive analytics domain, we expect more functionality in Tableau such as more predictive models and control in newly introduced forecasting feature and better integration with R (current Tableau version can do 4 functions passing parameters to R Server).
We will update this post as more information is revealed about Tableau 9.
[1] - Tableau's Building the 'Google for Data'
[2] - Tableau Gives Up Gains: Estimates Rise, Targets Decline, on Upbeat Q1, Forecast
[3] - Tableau Conference 2014 – A Field Report
[4] - Data Visualization with Tableau
No comments:
Post a Comment