Functional Practice

[tabby title=»Performed»]

1.1 Metadata documentation is developed, stored, and accessible.
An organization typically maintains information about data assets in multiple locations for di erent purposes. Metadata may be captured and stored
in data models, project documentation, operational documentation, or in business term lists. Collectively, these sources constitute a virtual metadata repository. Organizations should identify sources of existing metadata; evaluate their completeness, categorization, and properties; and plan to consolidate and enhance into a cohesive meta-model reflecting their needs.
As an example of existing metadata assets which may be leveraged, sources based on data models typically include the following items:

  • Entity type name
  • Attribute name
  • Table name
  • Column name
  • Data type
  • Length
  • Allowed values
  • efault values
  • Mandatory/Optional indicator
  • Data definitions for entity types, attributes, tables, and columns

Example Work Products

  • Metadata repository or virtual metadata repository (multiple sources)

[tabby title=»Managed»]

2.1 A metadata management process is established and followed.
An organization will benefit from defining a standard process for performing major steps in the metadata lifecycle (i.e., create, update, and delete). The process usually involves the following:

  • Identification of relevant stakeholders and their roles
  • Definition of data concepts, approved by the business
  • Determination of required metadata components and categories
  • Selection or building of a common repository for storage, maintenance, and retrieval
  • Configuration management and maintenance rules and criteria

Business stakeholder involvement ensures that metadata clearly describes information required for end users and supports the performance of critical processes in the data lifecycle, such as:

  • Data sourcing
  • Data movement
  • Targeting and classification
  • Usage (e.g., for reporting and within the system development life cycle— SDLC)
  • Governance and control

2.2 Metadata documentation captures data interdependencies.
It is important for the organization to determine the core set of metadata categories across the business, technical, and operational landscape to enable traceability from the sources of data to the target repositories.
The Data Management Lifecycle will help to identify where data are used and changed. This information helps to provide a source of information related to interdependencies.
Organizations often lack su cient metadata to construct a data lineage, which typically includes the following:

  • Data sources
  • Designation of authoritative or preferred sources
  • Applied transformations or aggregations
  • Data destinations
  • Integration mappings at the data element level showing how sources were combined when integrated
  • When new data is received
  • When data was changed, etc

For example, for a report provided to a regulatory organization, an organi- zation may need to trace the origin of the data from all sources through
any transformations, calculations, or aggregations applied from multiple processes, end to end. This provides a complete audit trail for exactly
what the data were, how they were changed, and how they were used in
the report. When data lineage is pieced together for a specific need, it is a time-consuming process that may be repeated in multiple contexts across the organization. The development of robust end-to-end metadata simplifies this process and facilitates its automation.

2.3 Metadata is developed and used to perform impact analysis on potential data changes.
Assessing the impact of data changes requires knowledge of the extent
of its use within the organization (i.e., by stakeholders, in application data stores, the data warehouse environment, etc.). In addition to metadata
that describes the data content, it is necessary to document the systems and stakeholders associated with each data set in the repository (through production, maintenance, and consumption). This may have an initial scope by data set, but the eventual objective is to achieve mapping at the attribute and column level. For example, when the business adds a new value to a highly shared type code (e.g., organization type), multiple data stores may need to modify their structures.
2.4 Metadata categories, properties, and standards are established and followed.
For each major classification of metadata (e.g., business, technical), the organization should propose and approve a set of categories (e.g., business terms, databases), and a set of corresponding properties (e.g., data steward, update date) for each category. Conventions for each metadata element representation should be specified. These decisions should be codified in a metadata standard(s) which is published, implemented, and followed.
Refer to Business Glossary, which provides more information about documenting and maintaining information about data categories and other metadata terms, etc.

Example Work Products

  • Metadata management policy
  • Business metadata
  • Metadata repository(ies)
  • Metadata Meta Model
  • Metadata governance and publication approval documentation (including business and technical stakeholders)
  • Metadata standards Audit results Metadata change log

[tabby title=»Defined»]

3.1 A metadata management strategy for the organization is established, promulgated, and maintained by data governance with input from relevant stakeholders.
Development of a comprehensive, complete, and well-organized metadata repository evolves over time. Similar to the recommended approach for the organization’s data quality program, the organization is advised to develop a strategy to collaborate on a top-down vision, goals, objectives, and to develop a roadmap for requirements, design, and development (or platform deployment)—which includes a phased implementation plan that accommodates dependencies and reconciles sequencing conflicts.
The metadata management strategy should address the following topics:

  • The organization’s shared vision for metadata management Stakeholder roles and responsibilities
  • Metadata attributes (extent, categories, properties, usage, etc.) Conceptual metamodel for the metadata assets
  • Data governance roles and responsibilities for metadata creation and changes
  • Compliance program for metadata management Sequence plan for guiding implementation

As systems become more integrated and coupled, and often include external systems where dependencies are based on contract, it’s important that
the metadata strategy addresses all of these data sources and related exchange standards. Adoption of externally accepted industry standards, as appropriate, helps to mitigate internal di erences and to increase the organi- zation’s interoperability with customers and clients, peers, and regulatory bodies.
Data Requirements Definition provides more information about developing data requirements with corresponding metadata.

3.2 The organization’s metadata repository is populated with additional categories and classifications of metadata according to a phased implemen- tation plan, and is linked to architecture layers.
The metadata repository may be built internally or purchased from third party vendors depending on the scope of requirements; in practice, it is usually preferable to buy rather than build, due to the cost of feature development versus feature purchase. Typical enhancements to the repository are subject to business priorities, the sequence plan, and available resources. The initial meta-model is enhanced and extended with model modifications to reflect increasing completeness. The following elements often comprise metadata expansion or extensions:

  • Data quality rules including rules related to source authenticity and provenance rules
  • Business rules
  • Extensible classifications (added by the organization)
  • Distinction between organizational and line of business classifications Authoritative data sources (trusted sources)
  • Data steward assignment to business terms
  • Data custodian assignments to physical data objects
  • Process metadata including transformation logic (e.g., ETL)
  • Linkage of the following architecture layers for data at rest:
  • – Business glossary
  • – Enterprise logical data model – Logical data model
  • – Physical data model
  • – Database schema
  • – BI layer
  • Physical data lineage (sources-transformations-targets)
  • Metadata history and versioning (enables audit and change control)

3.3 The data management function centralizes metadata management efforts and is overseen by data governance.
Metadata efforts include categorizations, properties, and standards; and organization-wide implementation.
Metadata differs from data as the range of issues is broader, more technical, and relates to other topics that are not usually part of the normal data lifecycle processes. Metadata is more highly impacted by projects, and its scope and depth make it challenging from a data governance perspective. Notwithstanding the important di erence between establishing versus maintaining metadata, aligning project-based metadata into an overall data governance activity is an important achievement, especially critical for master data management, data warehousing, and metadata repository implementation initiatives.
Due to the range and extent of metadata throughout the business and technology environment, roles and responsibilities will involve many sta
in multiple areas. It is usually best to assign the data management function to lead and drive the collaborative effort, and to fully inform and obtain approval from stakeholders, as a number of their staff members will need to devote time to the effort. Data governance, with representatives from across the organization, can address final approvals and ensure agreements.
Refer to Governance Management and Data Management Function, which provide more information about data governance structures and roles, and data management sta responsibilities.

3.4 Data governance approves metadata additions and changes.
Once a category of metadata has been released to the repository, additions
or changes to scope, categories, properties, and standards
should be managed by data governance and communicated across the organization.
3.5 Measures and metrics are used to evaluate the accuracy and adoption of metadata.
Key metrics and achievement targets should be identified in the metadata strategy. These can be further refined as the metadata assets are extended and expended.
Metrics define what needs to be tracked (and inherently convey why it is important); examples include the following:

  • The value of metadata management (link to cost containment, operating efficiency, process effectiveness)
  • The performance of key processes and procedures; for example, the frequency of use per attribute, how many data stores per attribute, how many applications per attribute
  • The criticality of data attributes to applications (which are core, for what process, used in which application, included in which calculation processes)
  • The costs associated with the movement of data across the lifecycle (and how much is at risk), especially if the lineage is fragmented
  • Tracking progress toward a single authoritative source
  • The quality of metadata breadth, depth, scope, availability, timeliness, accuracy, duplication, conformity, linkages, and clarity

3.6 Metadata, and any changes to metadata, are validated against the existing architecture.
For example, information about physical models, interface specifications, ETL scripts, data provisioning services, etc. should be validated to ensure that it correctly describes the existing environment.
Refer to Architectural Approach for more information about activities associated with data platform architecture.
Example Work Products

  • Example Work Products
  • Metadata management strategy
  • Metadata roles and responsibilities
  • Repository reports of metadata extensions
  • Metadata management organization standards
  • Metadata meta-model diagrams
  • Gap analysis results comparing implemented platforms against metadata Metadata metrics reports (adoption, percent complete, etc.)
  • Metadata sequence plan milestone and progress reports

[tabby title=»Measured»]
4.1 The organization has developed an integrated meta-model deployed across all platforms.
4.2 Metadata types and data definitions support consistent import, subscription,
and consumption practices.
Refer to Data Integration for information about generating and managing data mapping and transformation metadata.
4.3 The metadata repository extensions include exchange data representation standards used by the organization.
Examples of exchange data representation standards are XML and XBRL.
4.4 New metadata management activities are guided by metadata metrics and historical information about metadata.
4.5 Quantitative objectives guide metadata management and support process performance.
4.6 Statistical analysis reports for process, reporting, and performance are included in the metadata repository and employed to support fact-based decision making for new metadata management initiatives.
Example Work Products

  • Documented quantitative objectives for metadata
  • Documentation of measurement approaches to include the statistical and other quantitative techniques applied, and the performance thresholds
  • Comprehensive metadata repository reports including history
  • Metadata process e ciency reports
  • Unified or integrated meta-model
  • Metadata standards captured in meta-model
  • Documentation of uniform practices for import, consumption, and subscriptions of data
  • Documentation of standards followed for exchange data representation Reports of statistical results

[tabby title=»Optimized»]

Level 5: Optimized
5.1 Root cause analysis is conducted to reduce the variations between the repository information and the data it describes.

5.2 Performance prediction models guide changes in metadata management processes.
5.3 Quantitative metadata improvement objectives are derived from the metadata strategy.
5.4 Planned data changes are evaluated for impact on the metadata repository; and metadata capture, change, and refinement processes are continuously improved.
Example Work Products

  • Evidence of consistent and reporting based on standard definitions
  • Results of analysis of repository information against the data it describes
  • Documentation on prediction models
  • Define quantitative objectives
  • Impact analysis reports

[tabbyending]