An Introduction to Master Data Management (MDM) Presented by: Robert Quinn, Sr. Solutions Architect FYI Business Solutions
Agenda Introduction MDM Definition MDM Terms Best Practices Data Challenges MDM Benefits Technical Terms Q&A
Definition MDM is a set of processes and tools that enable an organization to define and manage non-transactional data entities
Non-transactional Data Entities Master Data = nouns (aka Domains, Entities) People (Customers, Employees, Vendors, etc.) Location (Addresses, Geography) Things (Products, Materials, Assets, etc.) Reference Data (Codes) Other (Date/Time, Chart of Accounts, etc.)
Practices, Processes and Tools Collect, Cleanse, Relate, Distribute, and Protect Standards Ownership / Approvals Understanding Communication
Big Umbrella Consolidate disparate data sources Matching / De-duplication Relationship Management Organizational Views Access Control / Ownership Cleansing / Data Quality / Standardization External Data Integration
Typical Data Challenges Multiple systems contain subsets of critical data Existing applications can t support new/different requirements Operational requirements conflict with analytical requirements Organizations can t agree on a single version of the truth Garbage in, garbage out
Benefits of MDM Collect Combine data from isolated systems to create a complete picture Cleanse Consistent / well structured data is easier to use Relate Relating entities together creates new information Distribute Consistency reduces confusion, promotes reuse and improves integration Protect Compliance and security are fundamental requirements
Recap MDM is a focused, organizational approach to business entities MDM is NOT concerned with the activity of the entities (orders, returns, loans, etc.) MDM should be more about process than technology Technology enables MDM, but it s the process that is MDM
Relationship Management Relating entities creates new information Matching is not just about duplicates House-holding Social Networks Shared Ownership Comparing relationships from multiple sources is a powerful discovery tool Effective dating provides a view of entities & relationships through time
Integrate multiple versions of the truth
Relationship Management Summary MDM is foremost about improving the consistency, quality and scope of your data Relationship Management is a process of generating new and valuable information from existing data Effective dating is the practice of tracking data changes through time
MDM Technical Terminology Operational vs. Analytical Single Domain vs. Multiple-Domains Federated vs. Centralized MDM Architectures Registry Transaction Co-existence Consolidated
Analytical MDM Architecture
Sources (Reference or Relationship Data) Staging MDM Hub ~ Data Flow MDM Trust Framework Rules-based Configuration Tools Delta Detection and Cleansing Consolidation Process Data Source ETL Landing Raw f(x) Reject Management Insert/ Update f(x) f(x) Apply Trust and Validation Queued for Merging Auto Merge Match Queued for Matching New Target Data Model Product Name Address Manual Merge Un-Merge Dynamic Cell-Level Survivorship Events Msg Queue Data Warehouse Data Consumer Application Msg Queue/ Services f(x) f(x) Insert/ Update Rules Metadata Content Delta Detection and Cleansing Hierarchy Validation State Mgmt Workflow Event Trigger History Lineage X Ref Trust Score Audit Application
Operational MDM Architecture
Architecture Recap There are numerous architectures, approaches, and toolsets for implementing MDM Architectural choices (as always) are dependent on your requirements, culture and long-term vision The technical landscape is maturing but should still be considered leading edge
Questions and Answers Thank you for your time! Robert Quinn FYI Business Solutions