Element Caches Overview and Best Practices

An element cache is a recently used attribute element list stored in memory for faster retrieval. When it hits a cache, a request for attribute elements runs against Intelligence Server memory instead of issuing a SQL statement against the database, thereby reducing response times significantly. Between 25% and 30% of all queries issued against the database in a typical BI implementation are element queries. This means that performance gains on element browsing will have a significant impact on overall system performance.

Element caches tend to be especially useful when users execute reports or documents that contain attribute element list prompts. The speed at which users see the options for elements in lists and thus, the overall usability of the report or dashboard is greatly enhanced by caching the elements present in the
prompt.

Element Caching Best Practices

Using Intelligent Cubes to Eliminate Element Database Queries
The execution of a typical prompted report sourced directly against the data warehouse generates two types of queries—element queries to populate the prompt and report queries to calculate the report results. On the other hand, a report sourced from an Intelligent Cube retrieves both the prompt elements and the report dataset directly from the cube, alleviating the load on the data warehouse and improving overall system performance.

Setting the Same Element Display Size for MicroStrategy Web and MicroStrategy Desktop
The amount of memory that element caches use on Intelligence Server (set at the project level) and on the MicroStrategy Desktop machine can be controlled. To avoid the risk of memory swapping and performance degradation, it is important to optimize the use of the memory assigned to element caches. If different element display sizes are set on Intelligence Server and MicroStrategy Desktop, the element caches cannot be shared between those two components,and the risk of memory swapping increases. Instead, defining identical element display sizes on both Intelligence Server and MicroStrategy Desktop allows element cache sharing and optimizes the use of a limited resource.

Assigning Enough Memory for Element Caching
As always, you should allocate sufficient memory in the server for the storage of element caches. Default values tend to be low. You can refer Microstrategy Installation guide for recommendations. You should consider allocating even more memory for cases when some attributes contain large amounts of elements.
An average element cache memory requirement is about 10 MB, for a display size of 50 elements. For 100 attributes, as seen in a typical implementation, it is 512 bytes per element.