Report Caches Memory Capacity Planning

Report caches are first created in memory within Intelligence Server. They are later backed up to disk and reloaded into memory when accessed. If Intelligence Server memory is insufficient to hold all new requested report caches, the least recently used caches are unloaded automatically from memory to make space for the new caches.

The process of loading a cache from disk can add significant delays to the user wait time. Large cache sizes take longer time to load and may displace other caches that will need to be moved back from memory to disk to make space. This process degrades performance both for the user who requested the cached report and for users who would have been able to hit the now displaced caches.

Based on the above considerations, from a performance perspective it is important to run MicroStrategy in a 64-bit environment whenever possible given the significantly larger amounts of addressable memory it offers. If this is not possible, you must at least ensure that all available Intelligence Server memory is used optimally.

Memory capacity planning for report caches can be extremely challenging, but at a high level, a good estimate of the amount of memory needed to effectively cache a given set of reports can be obtained as follows:

• Determine which reports should be cached on a given project. For example, you can use Enterprise Manager to determine the most frequently used shared reports as discussed earlier.
• Generate caches for those reports and record cache sizes (both binary and XML) using Cache Monitor. Note that in the case of prompted reports, you will need to generate caches for each prompt answer you wish to cache.
• If you plan to generate some user-specific report caches, you must add this variable into your total cache memory utilization estimates.
• After you have estimated the total amount of RAM that will be used for report caching you can define it in Intelligence Server

Practicing Cache Maintenance strategy is an important aspect of capacity planning for report caches and is necessary for a successful BI deployment. The list below contains recommendations for cache maintenance:

• Deleting Obsolete Caches—Establish a cache deletion/regeneration strategy to ensure users can access the most recent data available. Refer post
• Allocating Enough Disk Space for Caches—Ensure enough disk space is allocated to store all cache files. Otherwise, new caches will not be created. In addition, in a clustered environment, ensure that all nodes have the correct read and write permission to the location where caches reside.