JPivot: cube or ER schema

MBincole
2004-11-03
2013-04-29
  • MBincole

    MBincole - 2004-11-03

    Hi 2 evry1,

    I'm studying & working with Mondrian/JPivot.
    I've read all the documentation found in the releases, but I think I've misunderstood something...

    For example, I've noticed that JPivot takes too much time when it works with large DB.

    Does...

    1 ) JPivot read the data from a ER schema and then build a cube everytime the user invoke an operation ( slice&dice, drill ) or

    2) JPivot work on pre-existing cube ( created with Mondrian ) ?

    Probably, none of the above is correct...
    Please, make me know what you think about this.

    Thanks for reading

    MBincole

     
    • Sherman Wood

      Sherman Wood - 2004-11-03

      None of the above is correct....

      Mondrian is doing the work to query the database and do calculations. It has a cache to store results and uses this to improve performance above just hitting the database each time.

      Mondrian performance is a function of:

      * database size
      * database complexity: number of dimensions and facts
      * database configuration: indexes being used etc
      * complexity of MDX calculations being performed
      * size of Mondrian cache

      Sherman

       
      • MBincole

        MBincole - 2004-11-03

        So, do you assert correctly that

        Mondrian is only an "OLAP server" and
        JPivot is a kind of "graphic interface"

        Do you?

        MBincole

         
        • Sherman Wood

          Sherman Wood - 2004-11-03

          Yes

          Sherman

           
    • MBincole

      MBincole - 2004-11-05

      Mmm...

      Let me make an example.

      Think that you want to query a database using JPivot because you are interessed on sales per store and then interessed on sales per month. What happens?

      FIRST HP

      Does Mondrian build up an hyper-cube with all dimension?

      Then

      Does Mondrian send to JPivot a smaller cube based on passed parameters or does JPivot have got some libraries which let this program read the Mondrian hyper-cube?

      SECOND HP

      Does Mondrian build up a cube with the submitted parameters each time I make a new query?

      Then

      Is the cube send to JPivot or does JPivot read the cube?

      MBincole

       
      • Sherman Wood

        Sherman Wood - 2004-11-05

        >Does Mondrian build up an hyper-cube with all dimension?

        No.

        >Does Mondrian send to JPivot a smaller cube based on passed parameters or does JPivot have got some libraries which let this program read the Mondrian hyper-cube?

        also

        >Is the cube send to JPivot or does JPivot read the cube?

        Mondrian returns the results of MDX queries and metadata about the cube to JPivot as Java objects. JPivot uses the metadata to generate MDX.

        >Does Mondrian build up a cube with the submitted parameters each time I make a new query?

        No. Mondrian caches metadata about the cube and as many prior results as it can, given cache size limitations.

         
        • MBincole

          MBincole - 2004-11-05

          I think I've understood something about the data flow... but I'm a bit confused when you talk about metadata.

           
          • Sherman Wood

            Sherman Wood - 2004-11-05

            Metadata, in this case, is the dimensions, levels and measures that have been defined for the cube in the schema.

             

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks