You are on page 1of 2
Performance Tuning In QlikView Although Qlikview can analyze millions of rows of data (depends on your hardware), we have to ensure that we have developed the script in such a way that Qlikview can perform as you would expect it to do. The size and processing time of a QlikView document depends on the following factors: + Number of rows & columns-= it's really not necessary, we can load 5 years of data in a QlikView document instead of loading 10 years of data. We should be selective about what fields are to be included in the QlikView document and eliminate those fields which are currently not being used. It has been observed in one of our projects that by reducing the number of rows and columns, the QVD load time has been reduced from more than 1 hour to around 20 mins and the size of the application from 60 MB to 25 MB, + Number of Distinct Values Fields with distinct values use the most space. Using Count (Distinct) forces QlikView to switch from using multicore to use only one core. Timestamps with minutes / seconds can be truncated to store only date, thus reducing the processing + Data Model — QlikView can analyze multiple data sources but works most efficiently with “Star Schema’ or “Snowflake Schema’, The Star Schema consists of few fact tables (possibly only one) referencing any number of dimension tables + User Interface — |. Resource intensive expressions like Count (Distinct), If () should be avoided, Set Analysis should be used instead. I. Complex calculations should be done in script rather than in charts / tables IIL Macros should be replaced by built-in QlikView functions IV. Expressions which are used frequently in more than one chart / tables should be stored in a variable which reduces the calculation time. \V. Minimize the chart objects which are less important as chart expressions in minimized objects are not calculated, Other points which may help in better performance are: + Creating indexes may increase the performance of the QVD by around 30% + Rather than loading the entire data set, do an incremental load i.e. appending new data to the historic data, These are few performance tuning methods which may reduce the processing time to load the data and render the graphs. But these practices are not universal; apply these on a per situation basis. By Manisha Choudary Posted in: Glikview

You might also like