Hello All,
I have created a WEBI reports using Multiple Queries using Three Universe.Report is of two tabs but when i select the five days of data also it is taking very long time more than one hour to display the data.
Can you please let me know Is there any way i can make it faster?
Thank you
Pooja
ktm 
(BOB member since 2008-11-12)
Ask your DBA to take a look in to the Query.
Check if your universe has been updated recently and has any loops or contexts unresloved. make sure the Integrity is done and passed in the universe.
Try running the Raw query against the Database and compare the query run time and take it from there. Most likely it should be an Data load issue in the database , Index dropped , partition dropped, some thing like that. Work with your DBA better.
ThanksâŚ
JohnJustus (BOB member since 2007-06-25)
Just with reference to my old query that Jansiâs pointed you at and the FAQ that I was pointed to then too.
My issue at the time was that I was having to return loads and loads of rows of information in to a query and then perform a variety of tests on it in a report, filter and order it based on those tests and create a lot of variables (measures) giving me results like sum this where that is true, thatâs true thatâs true⌠Any application is going to struggle a bit crunching vast amounts of numbers into an A4 dashboard like that : I got round it a little, immediately at the time by fiddling with my Java control panel and increasing the amount of mem available to webi - something the IT bods here hadnât done.
In the long run, the full answer was to go through the list of actions suggested, work with our 3rd party supplier to get them to create measures & other variables I needed in their universes for me & the end result was a query & report that runs in seconds rather than looking at âprocessing requestâ for 10 minutes in my status bar!!
My question in the past was more on the lines of where the processing youâre seeing happening, actually occurs - on your pc or on the bo server. I think that the answer is âsomewhere between the twoâ! At the end of the day it doesnât really matter and I like to get the server/db doing as much work as possible before returning results to my reports for me to then play around with. 
SteveD 
(BOB member since 2009-11-02)
A good example of analysis and pushing the work to the appropriate engine level :).
Usually, there can be 3 engines involved, ETL, Query and report level engine.
âWorkâ should be pushed, upwards as far as possible. Ideally, a universe should have very little SQL functions in it apart from, sums, counts, prompts e.t.c. These will not be required if they are pre-canned by your ETL. This will also allow the maximum use of indexes, as you will have no SQL functions in query Where statements. Tables should be used, rather than derived tables or views, if using a standard RDBMS, to prevent contantly re-inventing the wheel, running code, at run time. The two platforms where, some, consider this not not so much of an issue, is parrallel processing "databases / appliances such as Terradata or Netezza.
At the query level, you should be looking at getting what you require, with aggregates in your queries e.t.c. to allow row aggregation to be dealt with by the database engine.
Finally, if you have done the above, you should require very little aggregation / projection to be done by the report engine, the weakest link if you like.
Averages, however, should be calcuated at the report level, to enable the, correct, evaluation of two phase calculations, when slicing and dicing.
Mak 1 (BOB member since 2005-01-06)
(BOB member since 2008-05-12)