TM1 Forum

Hello guys, im looking for the fastest way to load data into tm1 cubes.

I need to load 400000 records in one cube, numeric and string data.
i already have the excel file , but turbo integrator takes tooo long to load it.

so, what should be the best way to load this amount of data?
ODBC?
Package TM1?

Thanks for any tips.

Kazeh.

How long is too long?

400,000 records from any data source shouldn't take very long at all.

Are there rules on the destination cube? Is the load process incremental?


Standard things to do are turning off cube logging in the prolog (and back on in the epilog if required), destroy the rule in the prolog then re-add it in the epilog (not always beneficial but sometime is), generally consider how the TI is constructed and if the amount of calculations it does can be reduced.

Would be interested to see your TI and here more about your cube structure, that way people could probably give some tips.

The loading time for a cube usually depends more on the "write to cube" speed than the "read from source" speed.

One quick way to identify the cause of the delay is to create a dummy cube with the same dimensions as the regular cube to which you load. The dummy cube shouldnt have any rules.
If the load to this dummy cube takes the same time as the load to your original cube, the problem is very definitely on the "reading data from source" side.
If the dummy cube loads faster, then you should follow Declan's suggestions: set logging off, delete the ruleFile before the laod, and save the rule File again after the load is complete.
If there's no improvement, then its time for us to roll-up our sleeves and get to work on the optimization!

Yipee-kay-yaay!