Oracle Analytics Cloud and Server

Home› Oracle Analytics› Oracle Analytics Forums› Oracle Analytics Cloud and Server› Oracle Analytics Cloud and Server

Categories

Welcome to the Oracle Analytics Community: Please complete your User Profile and upload your Profile Picture

Join Subject Areas in 11g vs 12c

Received Response
21
Views
6
Comments
BIAP
BIAP Rank 4 - Community Specialist
edited Aug 13, 2024 5:58PM in Oracle Analytics Cloud and Server

Hi,

We a requirement to report on two subject areas. I have been doing some search online.

OBIEE TRAINING: querying across multiple subject areas in obiee 11g

Does the folder and column names (i.e. "customer detail"."customer number") have to exact match with the other subject area? Need to be exact same on logical tables as well?

Is there other consideration when joining subject areas?

Are there any significant improvements or new features in 12c comparing with 11g?

Thanks

Answers

  • Joel
    Joel Rank 8 - Analytics Strategist

    The most important thing to note about multi subject area reporting is that both subject area must be derived from the same logical business model.

    Your other question about the common fields. Well, from a user perspective, it would make sense for the columns to have a common name across subject areas although there is no technical reason for this as long as the joins exist between these columns in the model.

  • BIAP
    BIAP Rank 4 - Community Specialist

    Thanks for that Joel. The joins must exist for underlying physical tables? All the logical join will end up as physical join somehow as per my understanding.

  • Joel
    Joel Rank 8 - Analytics Strategist

    That’s correct. The joins are defined in the physical layer and the logical joins just define the type of - inner join,left outer, right outer, full outer for example.

  • [Deleted User]
    [Deleted User] Rank 2 - Community Beginner

    Cross-subject modeling isnt necessarily tied to literally "joining" things together in the physical layer. You cant "join" a cube and an xml file or a database and a Hadoop cluster. ...or just two simple tables residing in different databases.

    I think it would be good to also understand the concept of federated queries and in-memory stitch joins.

  • BIAP
    BIAP Rank 4 - Community Specialist

    Wow...You didn't answer my question, but you broaden my mind about cross-subject modelling...

    After some investigation, here are some tips for future readers:

    You can only report on measures from multiple facts where all dimensions that are used in the analytic are shared.

    Subject areas should have Common Dimensions.

    Add at least one metric from each subject area

    Turn on the enable dimensionality flag under advanced tab in the analysis.

  • [Deleted User]
    [Deleted User] Rank 2 - Community Beginner

    Yeah I am good at confusing people with even more questions.

    Regarding the "enable dimensionality" flag: i am not a fan because it more or less just hides incomplete RPD design and needs to be set in every analysis.

    Also as soon as you use anything else but Answers to access the RPD you'll have issues.