DMF Sample Questions & Answers

Accountability is a cornerstone of data ethics and data protection regulations like GDPR. It requires organizations to take responsibility for their data handling practices and to be able to demonstrate compliance with data protection principles. This includes having appropriate measures and records in place.

The ETL (Extract, Transform, Load) pattern involves extracting data from source systems, performing transformations in a separate staging area or engine, and then loading the already-transformed data into the target system. This is ideal for traditional data warehousing where complex business logic and data cleansing must be applied before the data is loaded into the structured target schema. ELT, by contrast, loads raw data into the target and uses the target system's power to perform transformations.

A Data Custodian is responsible for the technical environment and management of data assets. Their role is operational and focuses on tasks like implementing security controls, performing backups and recovery, and managing the database infrastructure. Defining business rules and approving access are typically responsibilities of a Data Steward or Data Owner.

A multi-model NoSQL database is designed to handle various data types and structures efficiently. For instance, a document database (a type of NoSQL) can store structured and semi-structured data within a flexible JSON/BSON format and can link to or store binary files (unstructured data). While an RDBMS can use BLOBs, it is not optimized for querying or managing the content of semi-structured or unstructured data, making a NoSQL approach more suitable for this use case.

Data modeling, especially the conceptual and logical phases, is a critical communication and analysis tool that ensures the data structures align with the actual business processes and rules. Skipping this step and allowing developers to create the schema ad-hoc often leads to a model that only serves immediate programming needs, failing to capture the true business context. This results in data that is difficult to integrate, report on, and adapt to future business changes.

Document and Content Management includes the sub-discipline of Records Management. Records Management is specifically concerned with controlling the creation, retention, and disposition of records according to legal, regulatory, and business requirements. Defining a seven-year retention period and subsequent destruction falls squarely within this area.

Survivorship is the process of resolving conflicting data values from different source systems to create the single, authoritative version for the master record. It involves applying predefined business rules (e.g., 'trust source system A over B', 'use the most recently updated value') to determine which attribute value 'survives' and becomes part of the golden record.

Online Analytical Processing (OLAP) is a technology that enables users to perform multi-dimensional analysis of data. OLAP cubes, which are often built on top of dimensional models, pre-aggregate data to allow for rapid 'slicing and dicing' by various dimensions (like region, product, time). This is precisely what is needed for the described interactive dashboard.

Data lineage is a specific component of metadata that documents the origin, movement, transformations, and destination of data. It provides a complete audit trail, which is essential for compliance, root cause analysis, and impact analysis. While it uses technical and operational metadata, the specific term for this end-to-end view is data lineage.

The root cause is the lack of a preventative control. A preventative control, such as an input validation rule in the application, is designed to stop data quality errors from occurring at the point of entry. Detective controls (like data profiling) find errors after they have occurred, and corrective controls (like data cleansing) fix them. The most effective approach is to prevent the error in the first place.