Re-Ordering of the dimensions to optmize my cubes

[dima55]

Hi All

I am using Re-Ordering of the dimensions to optmize my cubes. But i am stuck with the testing process.

I have selected 2 cubes in dev for my testing which shows -47% and -56% when i click on test in re-ordering feature. But i am asked to test the affect of this on reports. I am using TM1 persecpective version 9.5. When i click on browse cube and than calculate on particualr view it takes lot of time(before reordering). Will reordering show any significant changes in cube viewer which can be measured by any parameter ?. or Can i in persecptive anyway find the difference after reodering in terms of time it took to calculate/build the cube before Reordering and after reordering.

I just want to show the diffference done by reordering on these 2 cubes.

Thanks
Dima

[Wim Gielis]

No, you can only time the difference yourself.
Reordering dimensions will deal with memory consumption reductions or increases.

[rkaif]

As mentioned by Wim you will have to manually note the Time difference before and after of the Dimension re-ordering.

If you want to get some sophisticated results then you may use JMeter or IBM Rational Performance Tester.

[lotsaram]

Hi All

I am using Re-Ordering of the dimensions to optmize my cubes. But i am stuck with the testing process.

I have selected 2 cubes in dev for my testing which shows -47% and -56% when i click on test in re-ordering feature. But i am asked to test the affect of this on reports. I am using TM1 persecpective version 9.5. When i click on browse cube and than calculate on particualr view it takes lot of time(before reordering). Will reordering show any significant changes in cube viewer which can be measured by any parameter ?. or Can i in persecptive anyway find the difference after reodering in terms of time it took to calculate/build the cube before Reordering and after reordering.

I just want to show the diffference done by reordering on these 2 cubes.

Thanks
Dima

In terms of differences you can quantitatively show lower memory consumption which equates to more efficient hardware usage. For view retrieval performance you can likewise show quantitative changes but you have to measure it like Wim said - there isn't any way to know the exact effect on performance in advance as there are many variables in play however performance improvement is generally in line with the memory reduction but this isn't always the case the performance improvement could be many times more than the memory reduction or visa versa many times less, you simply have to measure it. Make sure when you do the comparative performance measurements that you start from a clean base, that is with the cube's cache cleared, so as not to distort the results and draw wrong conclusions.

[dima55]

Thanks alot Wim Gielis ,rkaif and lotsaram.

As i am new to TM1 and there is no one i know who work in TM1 this forum is helping me to learn a lot. Thanks to you all for your help.

Thanks
Dima