Question:
HI
IS there any big drawbacks of using too many key figures in the
cube, and if there is a big drawback in the same, then is it a
good idea to split the cube into two to accomodate the key
figures separately.
We have a cube which has about 60 key figures and we are planning
on adding more, can anyone throw some lights on the performance
implications if there are any, and if so, how do we tackle this one.
cheers
Answer:
As the key figures are in the fact table, I believe you should not encounter any drastic slow down on performance during query runtime.
At loading time, if you are calculating some of the key figures, you could have a problem of performance, but if you just upload it directly from the source system, no big issue.
Now, I just wonder in what scenario you are to have so many key figures... Maybe you could have less if you do the correct combination of characteristic and use restricted key figures in your reports, for instance ?
Just my two cents(*)
Ch
(*)euro-cents, of course, so more valuable than normal-cents
Answer:
Hi CHC
we are using this in Aggregated accounting, we are getting data for profitability analysis and also general ledger transcation figures, hence there are lots of calculations being done at each key figure,
wondering is using a start routine is better than using individual routines for each key figure,does that help in performance.
Answer:
Hi CHC
we are using this in Aggregated accounting, we are getting data for profitability analysis and also general ledger transcation figures, hence there are lots of calculations being done at each key figure,
wondering is using a start routine is better than using individual routines for each key figure,does that help in performance.
I though it would be something like this.
Ok, I don't tell you to change your modelling 'cause it would probably be a major change for you; but don't you think you could have limited the number of key figures by having characteristics for key figure origin, for instance ? (same concept than versions or value type in COPA for instance).
I've got the feeling that start routine would go faster, while I don't have any formal argument to prove this, because it all depends on what are the calculation you do (if you use twice the same input field for a calculation, you only have to read it once in a start routine, for instance).
Ch
Answer:
I didnt quite get the idea, of limiting the key figures by having characteristics for key figure origing ... can u please elaborate on that, it will be of great great help to me ...
cheers
Answer:
-brief explanation, as I don't know your scenario-
suppose that your cube is fed from FI and from CO, for instance;
suppose that you have KYF1 KYF2 KYF3 KYF4;
probably KYF1 and KYF2 are fed from a FI source, and on this line KYF3 and KYF4 are empty
probably KYF3 and KYF4 are fed from a CO source, and on this line KYF1 and KYF2 are empty
IF KYF1 & KYF3 are of the same type, and KYF2 & KYF4 are of the same type,
then you could have only KYF1 & KYF2 fed from both infosources, plus a characteristic CHAR1 fed with FI or CO;
in this case,
FI KYF1 KYF2 would be equivalent to KF1 KF2 and
CO KYF1 KYF2 would be equivalent to KF3 KF4
The former report that was running on KYF1 & KYF2 will still be running on KF1 & KF2, but with a restriction on CHAR1 = FI
the former report that was running on KYF3 & KYF4 will now be running on KF1 & KF2, with a restriction on CHAR1 = CO
(hope this is clear enough, please understand this is the end of my working day...)
Ch
Answer:
hey CHC
Brilliant .. that was a great suggestion ..
i am immediately gonna look up on this and see what i can
do ..
thanks a ton .. really appreciate it ..
Answer:
Just to give you some ease-of-mind : i'm currently in an environment where we have Cubes with nearly 200 Key Figures ..