Over the past years, several hub architectures for MDM have become the de facto industry standard. Each of the hub architectures offer pros and cons, and the following represents our views on each one:

The registry hub creates a MDM repository that typically consists of nothing but keys from other systems that contain the MDM data. Typically the data is not cleansed or transformed and is stored in a format similar to the data at the source.

  • Minimal changes required to source systems to integrate with MDM processes
  • The quickest of all hubs to implement
  • There is no continuous process of improving and governing data assets since the attributes of the data assets reside in the source
  • The performance of reporting queries becomes a factor of how many source systems that are keyed into the MDM since queries may have to be distributed across multiple systems to produce the desired result.
  • Data quality is still controlled only at the source
The repository hub creates a MDM repository that consist of a complete dataset of the enterprise data that has been identified as being part of the MDM. Typically the data is cleaned, transformed, and stored in a model that is different from the data at the source.

  • Maintains a continuous improving collection of data assets by moving the data assets under a data governance umbrella
  • Usually requires increased time and effort to implement

  • Challenges
    • Changes to applications that produce or consume the MDM data to integrate with MDM processes can be complex and challenging depending on the number and type of source systems
    • Developing a unified data model that will encompass all required data assets while keeping the model flexible and easy to maintain and consume
    The hybrid hub takes portions of both the repository and registry hubs and combines them. In a hybrid hub, the keys from the source system are still replicated into the hub and along with some of the attributes of the data that are most frequently queried.

    • Minimal changes required to source systems to integrate with MDM processesQuery performance, compares to only a registry, is improved by moving the most utilized attributes to the hub reducing the need to distribute queries across the source systems
    • A plan must be in place to ensure proper replication of changes in the source data attributes are accurately replicated to the hub
    No one architecture is better than the others. Ultimately it's up to the leaders in an organization to understand each hub architecture and determine which one will best support the organization's business and technology goals. Regardless of which approach you take, MDM is critical in establishing a consistent view of your enterprise data and ultimately creating "one version of the truth."

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.