Hi Guys,
I am currently implementing TM1 V 9.4.
In Tm1 only when the cube is browsed/Opened that the data and calculations of the slice are loaded into memory.
Is there any setting in Tm1.cfg wherein the data and calculations are automatically loaded onto the memory once I start the server?
So that when the user browses or opens the cube the cube opens faster as the data and calculations are already in memory.
Regards,
Girish
Loading data and Calculations into memory
-
gnampoothiry
- Posts: 14
- Joined: Fri Oct 24, 2008 9:18 am
-
ScottW
- Regular Participant
- Posts: 152
- Joined: Fri May 23, 2008 12:08 am
- OLAP Product: TM1 CX
- Version: 9.5 9.4.1 9.1.4 9.0 8.4
- Excel Version: 2003 2007
- Location: Melbourne, Australia
- Contact:
Re: Loading data and Calculations into memory
Unless the specific cube properties are set to "LoadOnDemand" then all cube data is loaded into memory when the database loads.
Consolidations are calculated on demand.
Once a view is loaded into memory the view is cached on the server and will be much faster for subsequent users to access. However, as all calculations are on demand if there are any underlying data changes that affect the view (that is any dimension at any hierarchy level) then the whole view would need to be rebuilt
The loading of views, especially those with many rule calculations or high level consolidations can be speeded up dramatically by pre-constructing them. However due to the reason given above this is only really practical in a "reporting cube" environment. That is it will work brilliantly if the reporting views only contain actuals from a source system but if you are dealing with a forecasting application that requires data input from users then this will be of no use as if there are any underlying data changes then any views would need rebuilding.
You can pre-construct cube views in TI with ViewConstruct.
There are also various config file parameters for memory allocated to caching views and optimising calculation of consolidations. I recommend you read the operations guide and developers guide.
Consolidations are calculated on demand.
Once a view is loaded into memory the view is cached on the server and will be much faster for subsequent users to access. However, as all calculations are on demand if there are any underlying data changes that affect the view (that is any dimension at any hierarchy level) then the whole view would need to be rebuilt
The loading of views, especially those with many rule calculations or high level consolidations can be speeded up dramatically by pre-constructing them. However due to the reason given above this is only really practical in a "reporting cube" environment. That is it will work brilliantly if the reporting views only contain actuals from a source system but if you are dealing with a forecasting application that requires data input from users then this will be of no use as if there are any underlying data changes then any views would need rebuilding.
You can pre-construct cube views in TI with ViewConstruct.
There are also various config file parameters for memory allocated to caching views and optimising calculation of consolidations. I recommend you read the operations guide and developers guide.