As explained in my another post, Visual insights dashboards in HTML 5 (MSTR 10) is the major change from Visual insights dashboard in Flash (MSTR 9.x)
Please refer the below link for more details:
In this post, I am giving few specific items on performance in HTML based Visual insights dashboards. Before get into the details, this may only applicable for projects with large schema. yes it is related with project schema size.
After moving to Microstrategy 10, we may see few differences in performances. In few cases, you will see better performance in MSTR 10 and in few cases, you may see bad performances in MSTR 10.
In this post, we are going to discuss about MSTR 10.x Visual insights dashboard performances and few tuning options
In Visual Insights, few activities where we may see slowdown in VI dashboard in MSTR 10
(1) Moving from one layout to another layout in VI dashboard
(2) Moving to different panels
(3) Applying different filter condition
basically most of the activities inside the VI dashboard. In HTML 5, for every single click, we are seeing the execution icon spinning for secs and some cases, it takes really long time.
Ex: Moving from one layout to another takes a minute - This is not very good
In general, as a nature of HTML 5, for every click or activity, it will execute and process and get the values. if it is within 2-3 seconds or max less than 10 secs then it is acceptable but anything more than that will create very bad user experience.
Below are few settings or configurations will help to improve the Visual Insights dashboard performance:
(1) Project default Drill Map:
a) Make sure you have a default project drill map configured in your project. In case it is empty then, MSTR considers/loads the huge chunk of your schema into the memory. You can see the same reflects in the size of your VI dashboard cache.
Execute the VI dashboard and check the size of your layout HTML cache. if you don't have default project drill map then you will see a big cache size and the same will impact the performance as well
b) In case your schema is very big with more number of attributes and facing performance issues in VI dashboards. Ex: Moving from one layout to another layout takes around 40-60 secs then Configure an empty drill map in project default drill map. This will reduce the cache size and give better performance.
(2) VLDB : MCE Join to get data combination:
Project Configuration -> Advanced -> Project Level VLDB ->Metrics
MCE Join to get data Combination : Select "Use Minimum relations to do the join with less memory"
This will set same behavior for multi cube engine as like MSTR 9.4.1.
This will improve performances in few cases especially when you use external data set and blending with other datasets in your dashboard.
Note: Please double check the data
(3) Join behaviour:
In VI dashboard, go to File ->Dashboard Properties -> Join Behavior -> Uncheck "Allow joins across datasets based on unrelated common attributes"
This option will help to improve the performance in few cases.
Note: Please double check the data
(4) Cube Footprint:
In case you are using Intelligence cube based on external source data (File or external data connection) and seeing performance issue or memory error then please check the cube memory foot print. Compare your MSTR 10 cube foot print with MSTR 941 cube memory foot print.
For sure you will see some difference. If the difference is within acceptable range then it is fine but in case you see a huge increase in MSTR 10 then you have to consider the below points.
a) Check your schema size. If your project is having huge number of attributes then mostly you will see this large cube memory footprint. In MSTR 10.x, It considers your whole project schema and keep that in cube memory to improve the blending performance but in case of large schema with small and medium size cube, this will impact the memory and performance drastically.
We don't have any solution available at this point. We can expect some improvements in the future microstrategy releases.
At this point, below are the different options (few options may not be feasible for all environments),
a) Reduce the project schema size if possible. Reduce the number of attributes
b) Move the respective dashboards and its dependents alone to different new project
c) Increase the server memory and memory allocated to Intelligent Cube
In case we have project with large schema especially with more number of attributes,you will see an impact in cube memory foot print and VI dashboard performance in MSTR 10 . Few of the above mentioned configurations may help in few scenarios. Anyway we will get the best performance with lean schema in MSTR 10.x