TM1 Forum

Hello, I have two cubes with an account dimension of 9000 accounts ( I am not forecasting all, probably 120), I am trying to copy a forecast version to another cube. I am doing balance sheet and income statement relationship between them, that is in certain way complex.

The view construct process is about 45 seconds for each consolidated level (top levels) on each cube. Why it takes so long information to be copied from one formulated cube to a another cube. TM1top reports going back to the same dimensions and cubes so and so and so... i am on 45 minutes and it is not finishing. Anyone faced a situation like this?

Regards,

Usually when the view performs well enough but the process takes orders of magnitude longer than expected, it can be attributed to rules/feeders. If there are rules in your destination cube, and especially conditional feeders, then they may be recalculated over and over as each individual datum is written to the cube.

If this is your problem there is an article on the old Applix Recommended Practices site (still there!) at http://www.applix.com/rp/Unloading_Rule ... rocess.htm that describes how to diagnose the situation and work around it. (May require an old Applix icc login. I do not know whether this tip sheet has been copied anywhere into the Cognos site.)

If it does not look like rules are the culprit, you could describe a litle more about your model and someone arround here will probably spot the problem.

What version are you using?

I agree with Mike. Test it out by creating a clone of your destination cube without rules and importing the data into that. If it's very quick then you need to use TI to remove and reapply the rules. Fiddly but well worth it - I cut a version copy routine down from 26 hours to 20 minutes this way.

Thanks guys,

I am using 9.1.2.U3

I believe the problem is on this cashflow and interests after them and the use of string formulas (STR(NUMBR(ATTRS) to connect both cubes I am making some tests.I ## some rules and time seems to be faster but am not sure why is so slow, i believe my response time are not so bad. I hope this description works for help.

(STR(NUMBR(ATTRS)

FYI using either the STR() & NUMBR() will eat up a lot of memory. Also why would you convert a string to a number and back to a string?

I was avoiding the use of many attributes.. i am not using many of that cases, on my previous answer i was lost where to start.

I am solving the problem thanks to "conditional feeders can be guilty" (thanks mike!), so I am reviewing other possible flaws with them. I notice that a trace feeder delayed for many minutes, and I started attacking that point, now my version copy is about ten seconds (# some feeders or formulas), but I need to continue checking this conditional feeders issue.

This feeders theme gets interesting on every step