Master Data Management
Master data management keeps shared records about core business entities, such as customers, products, suppliers, and locations, consistent across systems. It combines governance, data quality, matching, and distribution so teams can use trusted entity data.
itEnterprise architecture and integration | OpenSkills.info
Intro
Master Data Management
Master data management, or MDM, keeps the shared description of a business entity consistent across systems. An entity can be a customer, product, supplier, location, employee, or another subject that many processes reuse.
Picture one supplier stored in purchasing, finance, logistics, and analytics systems. Each system may use a different identifier. Names and addresses may differ. One system may hold the newest payment term while another owns the approved tax identifier.
MDM gives the organization a governed way to identify those records, resolve conflicts, and share trusted entity data. It joins people, policy, process, and technology. A database alone does not do the job.
Master data and its neighbors
Master data describes the durable entities around which business activity occurs. Transaction data records events involving those entities. An order is a transaction. The customer and product on the order are master data.
Reference data supplies controlled values used to classify other data. Country codes, currency codes, and status values are common examples. Metadata describes data, such as an attribute definition, owner, or source.
These categories can overlap in practice. The useful question is not which label wins. Ask whether many systems depend on a stable identity and shared meaning for the entity.
The managed entity
Continue the course
This section is part of the paid course.
See pricing to subscribe, or log in if you already have access.
Sources
- https://www.ibm.com/think/topics/master-data-management
Supports
- MDM combines technology, tools, processes, data models, and stewardship across enterprise systems
- Master data describes core entities such as customers, products, suppliers, and locations
- MDM work includes integration, standardization, matching, deduplication, reconciliation, enrichment, governance, and distribution
- Golden records integrate source records into a unified entity view
- Master data differs from transaction data, reference data, metadata, and other organizational data
- Recurring source inconsistency, quality problems, and synchronization are central MDM challenges
- https://www.ibm.com/docs/en/ws-and-kc?topic=data-concepts
Supports
- MDM creates entities by matching records from one or more data assets
- Source-system identifiers track record origin
- MDM data models include record, attribute, relationship, hierarchy, and entity concepts
- https://www.iso.org/standard/62392.html
Supports
- ISO 8000-100 covers fundamentals and requirements for master-data quality
- The ISO 8000 master-data series includes quality management and quality metrics
- The standard addresses computer-checkable master-data exchange at system interfaces
- https://dama.org/learning-resources/dama-data-management-body-of-knowledge-dmbok/
Supports
- DAMA-DMBOK places master and reference data within the wider data-management discipline
- The body of knowledge connects governance, quality, architecture, modeling, integration, metadata, and related functions
- https://help.sap.com/docs/SAP_S4HANA_ON-PREMISE/6d52de87aa0d4fb6a90924720a5b0549/56a57357f2b1aa6be10000000a4450e5.html
Supports
- Central governance controls creation and change through workflow, validation, approval, activation, and distribution
- Consolidation loads distributed source data, standardizes it, detects duplicates, and calculates a best record through survivorship rules
- Consolidation and central governance can be combined in a coexistence approach
- Best-record decisions depend on duplicate groups and attribute-level survivorship rules
- https://learning.sap.com/courses/sap-master-data-governance-on-sap-s-4hana/explaining-consolidation-in-master-data-governance
Supports
- SAP documents central governance, decentralized consolidation, data-quality management, and coexistence as MDM approaches
- A consolidation flow can load, check, standardize, match, calculate best records, validate, and activate
- Consolidation can support analytics, initial loads, mergers, and mixed ownership
- https://learn.microsoft.com/en-us/purview/data-governance-master-data-management-semarchy
Supports
- A representative MDM architecture includes ingestion, quality, validation, matching, deduplication, authoring, curation, and collaboration
- MDM can integrate governed metadata with a cloud data platform and operational sources
- https://learn.microsoft.com/en-us/openspecs/sql_data_portability/ms-dpmds/8eb2e6be-e4f1-4ecc-9486-50fc0592bb75
Supports
- MDM supports shared definitions, stewardship, business rules, workflows, hierarchies, and publication
- Critical entity domains can include product, customer, location, equipment, employee, and vendor
- https://github.com/sindresorhus/awesome
Supports
- The Awesome master list provides the starting discovery path for data-engineering ecosystem lists
- https://github.com/OlivierBinette/Awesome-Entity-Resolution
Supports
- The focused list curates Splink, Zingg, and dedupe as end-to-end entity-resolution software
- The list distinguishes entity-resolution tools from supporting comparison, cleaning, quality, and blocking tools
- https://moj-analytical-services.github.io/splink/index.html
Supports
- Splink performs probabilistic record linkage for deduplication and cross-dataset linking
- Its documentation covers blocking, comparison, training, prediction, clustering, diagnostics, evaluation, and several SQL backends
- Match evaluation uses labeled evidence and exposes tradeoffs between linkage errors
- https://docs.zingg.ai/latest
Supports
- Zingg provides entity resolution for building unified core business entities
- Zingg uses labeled pairs and active learning in its entity-resolution flow
- https://docs.dedupe.io/en/stable/
Supports
- dedupe is a Python library for structured-data deduplication and entity resolution
- dedupe learns weights and blocking rules from human-labeled data