Primecube

Make your dossier run faster in 5 steps

Dossier performance is still one of the most asked requests we receive from our customers. We dug into details and created some time ago detailed articles about how to improve it and divided it into three parts:

The tips are still correct and worth trying out if the main steps that we will propose here will not work. These points are a quick way to make your dossier faster in a short time. So let us begin!


1.Is your dossier fast without any visualizations on it?

Try to remove all the dossier content so that there is nothing to render on the screen and run the dossier. If it does not meet your SLS’s then this means that we have a problem with one of the underlying dataset. Try to find it using a binary search and think about improving its performance. If your dossier is still slow, please move on to step number 2.


2. Purge, purge and purge

Delete all not used derived elements and datasets from the dossier!


3.  Try selecting option to load chapter on demand (if your dossier loading time is slow)

This option is available in File > Dossier Properties from the top menu bar.

[/et_pb_text][et_pb_text admin_label=”Imagen” _builder_version=”4.7.3″]

If the dossier loads fast then this means that one of the chapters renders slowly and we need to analyze it to see why. If it still does not meet your SLSs try saving dossier and run the dossier on other chapters to find the slow one(s). Or do the same binary search (by deleting one by one) to find in which chapter we have a bottleneck.


4. Once the chapter or chapters are identified, see which visualizations are slowing your dossier down

From the experience I always look at the total render time. I saw that this is not equivalent with the real render time and we need to take a closer look on visualizations that take longer than 1 second to render. What can be done to improve their performance? Here are the quick tricks:

  • Does the visualization have a filter on it?

If yes, try to filter the metric itself instead of whole visualization. Important – filter metric using the case statement and try avoid using filtered metrics.

  • Does the visualization uses filtered metrics?

Try converting them into the case metrics.

  • Does visualization uses derived metrics?

Try converting them into the standalone metrics. If not possible, try to convert them into derived metrics in the view report (if this is the case)

If non of the above worked this is the main trick that might work: try using the aggregated dataset to render the visualization.  Imagine that the dataset that feeds visualization with data is on the day level, but we display month aggregation on the visualization. Try creating another dataset on month level and then appoint it to the visualization. This trick will only work if we do not add a heavy dataset into the dossier or on that creates heavy virtual joins. This need to be tested but this hint worked for us when we had a slow visualization problem.

If non of the above works maybe you are trying to render to much data in visualization that is not a grid? Grids allow you to take advantage from incremental fetch and are perfect to display detailed and massive data.


5. For slow filtering please remember about the option to apply filter selection when all the filters are already chosen. “All at Once” option.

These are the basic rules on how to quickly improve your dossier speed. If non of them have worked, please go through our detailed publications on improving the dossier performance.