3 Replies Latest reply: Nov 1, 2012 1:04 PM by Raj Vadde RSS

    Performance issue

    Raj Vadde

      Hello ,

            

      I'm having a major performance issue with my application which is around 250 Mb, with 32 M records and  50 columns .

       

      The major issue is ..we have set of filter and object in given view when a user tries to  select few filters ..for each filter select the view is geting calculated and

       

      it won't allow user to select other filters  in this mean time and also for second filter select it again getting calculated and taking a lot of time.

       

      this should be reduced. i'm working this on a server of 24GB RAM.

       

       

      I have few more question regarding performance

       

      1. conditions in layout or condition in expression (i'm using QV11) which is better for performance.

       

      eg. straight table with 6 expression only 2 expression can be changed by giving condition instead of having a new straight table with those 4 old + 2 new exprssion. which is better for performance

       

      2. i have 3 qvd i'm force concattenting  them at  app level (all are optimized load)

       

      3. or concatenate 3 qvd to one major qvd  and use that in app (with optimized load)

       

      4. will loading time has impact on calculation time of object in the view.

       

      5.my data model is 1 major table with 49 column and 3 column - maybe spilting the (main table) is good idea to enhace the performance.

       

      Regarding data model is it fine to have all distinct fields in one table (one qvd)  in my case it less then 10 felds (most facts and majorly dimesions used at object level).

       

      all fileds with less num of distinct vaules in other tables .

       

      say suppose i have a field of zone (1k distinct values), sales (1K distinct values), division(4 vaules), income range(4 values) , age (4 vaules),  children(2 values)zonetype(2 values)

       

      in the above case. is it advisable to have zone and sales in one table  and other fileds in other table . or every thing in one table

       

      but most my object run by division filed ( and other fields like -- income range , age ,children  are used as filters.)

       

      which is the best ways to optimize my application ..and i'm using mostly set anaysis in my expression and i don't have any date fields.

       

      seems to be pretty long list but wil be very helpful if you guys provide your thoughts for these conditions

       

      Thanks

      Raj

       

       

       

       

      in case as object (straight table i'm running them by division as dimension).

        • Re: Performance issue
          Gysbert Wassenaar

          This discussion has a lot of tips to improve performance. I hope that helps give you some pointers/ideas.

          • Re: Performance issue
            Raj Vadde

            5.my data model is 1 major table with 49 column and 2 table with - 3 column - maybe spilting the (main table)

             

            is good idea to enhace the performance ?

             

            Regarding data model is it fine to have all distinct fields in one table (one qvd)  in my case it less then 10 felds (most facts and majorly dimesions used at object level).

             

            all fileds with less num of distinct vaules in other tables .

             

            say suppose i have a field of zone (1k distinct values), sales (1K distinct values), division(4 vaules), income range(4 values) , age (4 vaules),  children(2 values)zonetype(2 values)

             

            in the above case. is it advisable to have zone and sales in one table  and other fileds in other table . or every thing in one table

             

            but most my object run by division filed ( and other fields like -- income range , age ,children  are used as filters.)

             

             

            in this case the  object is straight table with "division"  as dimension).