In the context of data mesh, "data products" refer to self-serve, domain-oriented data assets that encapsulate a specific business capability.The concept of data mesh, popularized by Zhamak Dehghani, emphasizes decentralizing data ownership and treating data as a product.

By adopting the principles of data mesh, organizations aim to overcome challenges related to data silos, centralization, and scalability, ultimately improving the agility and effectiveness of their data capabilities.

5 steps to configure Data Products in Collibra

  1. Define Data Products

A business intelligence report may be classified as a data product. Figure 1 shows a PowerBI-based Inflation report in Collibra. The report also includes a custom attribute for the value that has been assigned to the report.

Figure 1: PowerBI report in Collibra

  1. Develop Derived Data Products

Derived data products are based on one or more data products. For example, a bank has an Office of the Comptroller of the Currency (OCC) Compliance data product, which is derived from the Report Certification data product, which is derived from several report data products including Inflation (see Figure 2).

Figure 2: Banking data products relating to OCC compliance and report certification

  1. Establish Data Quality Scores for Data Products

The data quality score from Collibra OwlDQ may be appended as a custom attribute to the data product. For example, the data quality index for the Customer Master data product is 80 percent (see Figure 3).

Figure 3: Customer Master data product has a data quality score of 80 percent

  1. Agree on Data Privacy Rules for Data Products

Data products need to be constrained by data privacy rules. For example, the Chat Log data product is constrained by a rule that highly confidential data such as Social Security Number must be masked before usage for analytics (see Figure 4).

Figure 4: Privacy rule for chat log data product

  1. Create Customer-Specific Contractual Rules for Data Products

Data products also need to be constrained by contractual rules that are specific to customers. For example, an investment manager may have customer contracts that permit differing levels of data usage. The investment manager’s contracts with IBM and Honeywell only permit usage of 401(k) participant data within the app. However, the contract with Information Asset does not have any such usage restrictions (see Figure 5).

Figure 5: Mapping of data products to privacy and contractual rules