242 IBM Cognos Dynamic Cubes
8.7 Cache size best practices
For more information about suggested best practices when estimating cache sizes for a cube,
see Understanding Hardware Requirements for Cognos Dynamic Cubes at the business
analytics proven practices website:
https://www.ibm.com/developerworks/analytics/practices.html
The section about detailed sizing recommendations describes how to compute the minimum
size requirements for each of the caches and describes other considerations in determining
sizes for the caches. In general, having more CPU cores and allotting more memory for the
data cache can improve performance of Cognos Dynamic Cubes.
8.8 Scenarios for performance tuning
For certain scenarios, performance tuning can be used in caches.
8.8.1 Cases for adjusting cache sizes
The amount of memory per cube is the total of that required for the following items:
? Member cache
? Data cache
? Aggregate cache
? Temporary query execution space
? Additional space to reduce occurrence and cost of JVM garbage collection
This section describes cases in which the amount of memory that is associated with these
elements necessitates adjustment of the cache sizes of the cube and possibly the overall
QueryService JVM heap size.
Actual number of members is significantly larger than the estimated
number of members
For information about estimating cache sizes for a cube, see Understanding Hardware
Requirements for Cognos Dynamic Cubes at the business analytics proven practices
website:
https://www.ibm.com/developerworks/analytics/practices.html
One approach to estimating cache sizes is to compute based on the size of the two largest
dimensions of the cube.
Typically, the two largest dimensions of a cube are larger than all other dimensions by orders
of magnitude. If this case does not apply to your situation, use the actual number of members
across all the large dimensions, or even all the dimensions, of the cube to calculate the size of
the member cache. Then, use the new member cache value to recompute the other cache
values and overall memory for the cube.
Use these same steps to recompute the member cache and other cache sizes if the
dimensional space changes or grows significantly.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.