Hitachi Vantara Pentaho Community Forums
Results 1 to 6 of 6

Thread: OLTP tables VS RDBMS tables - Designing OLAP cubes - Peformance factors

  1. #1
    Join Date
    Feb 2013
    Posts
    530

    Default OLTP tables VS RDBMS tables - Designing OLAP cubes - Peformance factors

    Hello forum,

    Mondrian ol documentation is suggesting that there we should have star schema based RDBMS to develop multi-dimentional database( OLAP cubes).. Is it mandatory to have a warehouse in start/snow flake schema ?

    1) If not, what are the performance factors associated with it ?
    Here we have oracle database with normalized sachem's.

    Any suggestions on it will be appreciated.

    Thank you.
    BI developer
    Sadakar

  2. #2
    Join Date
    Jan 2013
    Posts
    796

    Default

    Not mandatory, but there are certainly negative performance implications to violating star schema best practices. And from a conceptual level it's a lot easier to think about cubes/dimensions/measures when they map directly to database structures.

  3. #3
    Join Date
    Nov 2008
    Posts
    777

    Default

    To expand on what mcampbell wrote, you basically have either a star schema or a snowflake schema. If you have a dimensional database model (described in depth in the writings of Ralph Kimball, et al) then you will have a star schema. If you have a normalized database model then you will have a snowflake schema. Either type will work but a star schema provides simplicity (as mcampbell wrote) and efficiency (much fewer join queries) at the expense of larger non-normalized database tables.
    pdi-ce-4.4.0-stable
    Java 1.7 (64 bit)
    MySQL 5.6 (64 bit)
    Windows 7 (64 bit)

  4. #4
    Join Date
    Feb 2013
    Posts
    530

    Default

    Thank you .

    I used to work with MySQL, Postgres where I had star-schema based warehouse earlier and developed CUBES and analysand.

    Now, I am working on oracle.. In oracle we have one database(SID) and different users(Schemas).

    Basically, it's an OLTP database(huge amount of data) and my organization is not focused on ETL to convert into start/snowflake schema at present.(In future we will implement it to create warehouse).

    Now, I need to show a demo on Analysis part using this databases.

    1) I have take 2 dimension fields coming from a table in one schema (lets say : SID : orcl and schema name: schema1)
    2) I have to take 2 fact fields coming from another table of another schema ( lets say : SID : orcl and schema name : schema2).

    How can I do this ?

    B'z schema workbench allows to connect to only one user at a time with any database.. in this case only with one schema ..

    Any inputs and suggestions ? What can I suggest to my highers on this ?

    Thanks
    Sadakar
    BI developer.

  5. #5

    Default

    I guess your best bet would be to simply create a new schema and use it to model star schema getting the dimension and fact fields from respective oracle schema.

  6. #6
    Join Date
    Feb 2013
    Posts
    530

    Default

    Thank you guys...!!!

    Worked with hard coded values of synonyms

    Not the best way of doing , but needed to demonstrated.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
Privacy Policy | Legal Notices | Safe Harbor Privacy Policy

Copyright © 2005 - 2017 Pentaho Corporation. All Rights Reserved.