Join the social network of Tech Nerds, increase skill rank, get work, manage projects...
  • Introduction to Master Data Management

    • 0
    • 1
    • 0
    • 0
    • 0
    • 0
    • 0
    • 0
    • 2.55k
    Comment on it


    This tutorial will guide you about Master Data Management Hub, Informatica. It is intended to provide Project Stakeholders, Business Users, Business Analyst and Technical Teams with the technical design and configuration of the
    1. Landing tables
    2. Staging Tables
    3. Staging Process Mappings
    4. Base Objects.

    In addition, this tutorial provides the source-to-target mapping of data elements between the tables that are part of the staging process.

    Data from Source(s) is captured and Loaded into MDM via import SQL Script/CSV import method using SQL Developer to Landing tables.

    MDM Process starts from Landing tables and runs each row of each Landing table through its corresponding Stage Map. Records that pass the Stage maps are Inserted into Stage Tables corresponding to each Source and Base Object combination. Then the Load process picks the data from Staging tables and inserts into Base Objects. Once all the sources are loaded in the Base Object, Tokenization process runs followed by Match and Merge. Once these processes complete, Records are consolidated and surviving records are termed Golden.

    Landing Tables

    Landing tables represent the data tables where data provider will introduce better quality data for MDM processing. To identify the source system for the data, each contains a Source System column within the table.

    Staging Tables

    Staging tables are derived from a combination of the base object tables and landing tables. The columns within the staging tables will mimic that of the base object tables. However, the data to populate the staging tables will come from the landing tables.

    It is supposed to exist one staging table for each source system that has information to load in the Base Object.

    Staging Process Mappings

    The logic to load data into the Staging Tables, coming from the Landing Tables, is designed into the Mappings. For each mapping, one (and only) Landing Table is read and the rules to load data into the Staging table are applied. We have one Staging Table for each Source System, for each Base Object.

    Base Objects

    In MDM Hub, central business entities, such as party, party address, party phone, party email and lookup reference values are represented in tables called Base Objects. A base object is a table in the Hub Store that contains collections of data about individual entities. Each individual entity has a single master record, or the best version of the truth, for that entity. An individual entity might have additional records in the base object that contain the multiple versions of the truth that need to be consolidated into the master record. Consolidation is the process of merging duplicate records into a single record that contains the most reliable cell values from all of the source records.

    The load of Base Objects happens through the Staging Tables. So, for example, if two source systems have data to be loaded into a given Base Object, there will be two Staging Tables associated to that base Object. The data always come from a Landing Table and is mapped to the Staging Tables into the Staging Process Mappings.

    A single Base Object has several internal/system tables to collect rejected data, keep historical data, control the loads, keep cross-reference data, and so on.

    After the load of the Base Object, a tokenize process starts. This process generates match tokens that will be used subsequently by the match process to identify candidate Base Object records for matching.

    Match/Merge Setup

    The match process begins, based on match rules defined by the user and configured in the Hub. The comparisons between records can be exact or fuzzy. The results of the match process can be an "auto merge" or a "manual merge" of the records.

    And, when the match is done, and the merge is ready to be applied, it appears the concept of Trust. In few words, the most reliable information among the source systems will be elected to appear in the Golden Record. You can configure some other trust-related variables, such as minimum and maximum trust, type of trust decay curve and how the time affects this curve.

    At the end, of Merge the records in the Base Object will be the "Best Version of Truth"

 0 Comment(s)

Sign In

Sign up using

Forgot Password
Fill out the form below and instructions to reset your password will be emailed to you:
Reset Password
Fill out the form below and reset your password: