Key Points
References
| Reference_description_with_linked_URLs_______________________ | Notes______________________________________________________________ |
|---|---|
data-governance-notes1. gdoc. link | |
What_is_Data_Governance_and_Why_Does_it_Matter_updated.pdf link What_is_Data_Governance_and_Why_Does_it_Matter_updated.pdf. file | |
dbx_data-and-ai-governance-6sept2023.pdf. - databricks link dbx_data-and-ai-governance-6sept2023.pdf. - databricks file | |
Key Concepts
What_is_Data_Governance_and_Why_Does_it_Matter_updated.pdf link
Data governance (DG) is the process of managing the availability, usability, integrity and
security of the data in enterprise systems, based on internal data standards and policies that
also control data usage. Effective data governance ensures that data is consistent and
trustworthy and doesn't get misused. It's increasingly critical as organizations face new data
privacy regulations and rely more and more on data analytics to help optimize operations
and drive business decision-making.
A well-designed data governance program typically includes a governance team, a steering
committee that acts as the governing body, and a group of data stewards. They work
together to create the standards and policies for governing data, as well as implementation
and enforcement procedures that are primarily carried out by the data stewards. Ideally,
executives and other representatives from an organization's business operations take part,
in addition to the IT and data management teams.
While data governance is a core component of an overall data management strategy,
organizations need to focus on the expected business benefits of a governance program for
it to be successful,
Data Governance problems
Without effective data governance, data inconsistencies in different systems across an organization might not get resolved. For example, customer names may be listed differently in sales, logistics and customer service systems. That could complicate data integration efforts and create data integrity issues that affect the accuracy of business intelligence (BI), enterprise reporting and analytics applications. In addition, data errors might not be identified and fixed, further affecting BI and analytics accuracy.
Data Governance programs
compliance with data privacy and protection laws, such as the European Union's GDPR and the California Consumer Privacy Act
(CCPA). An enterprise data governance program typically includes the development of
common data definitions and standard data formats that are applied in all business systems,
boosting data consistency for both business and compliance uses.
<<key < FACTUR3DT.IO - measure Data Governance Value in the organization against defined OKRs, KPIs in the Data Governance Program
<<key < key benefits possible include
setting, implementating and enforcing well defined data governance policies
data access controls and management
data quality management against internal standards and external regulations
compliance with data management policies and regulations
easy data availability for users and services
automation of data management tasks
data valuation based on usage vs objectives
compliance with data usage controls - PII, GDPR, CCPA etc industry data controls PCI HIPAA etc
Data Governance Mgt -
Management team sets governance rules with Board approval
Data Governance committee enforces the rules and approves all policies
CDO - Chief Data Officer role administers data governance processes
Data stewards own and maintain their responsible data sets and stores
EA sets the data management architecture
Data Governance Components
mission statement for the program, its goals and how its success will be measured, as well as decision-making responsibilities and accountability for the various functions that will be part of the program that is published
Data Governance Tools
data governance software can be used to automate aspects of managing a governance program. While data governance tools aren't a mandatory framework component, they support program and workflow management, collaboration, development of governance policies, process documentation, the creation of data catalogs and other functions. They can also be used in conjunction with data quality, metadata management and master data management (MDM) tools.
Data Governance Program
steps to take, including the following to-do items:
identify data assets and existing informal governance processes;
increase the data literacy and skills of end users; and
decide how to measure the success of a governance program.
- identify and update existing data management policies to meet program objectives, compliance and related regulations
- identify and create the appropriate metrics, data controls and audit procedures for the policies
- setup the data governance team and responsibilities including data stewards
- define the tools needed for each stakeholder to effectively manage their governance responsibilities
- setup data classifications to support data policies and access controls ( AATR etc )
- define common business meta data models and glossaries that map to the industry, the business and related standards
- build and maintain the enterprise data catalog with all internal and external sources, related controls for quality & currency,
- define all clients for the catalog resources and their related responsibilities and access to data catalog sources
How to Build a Data Catalog - tech target article
Data Catalog: A Comprehensive Guide
- Smith, Anne Marie, Ph.D. (2022). "How to build a data catalog: 10 key steps."
Table of Contents
- Introduction
- Definition and Importance
- Why Data Catalogs are Essential
- 10 Steps to Building a Data Catalog
- 3.1. Document Metadata Management Value
- 3.2. Identify Data Stewardship Uses
- 3.3. Design a Subject Area Model
- 3.4. Build a Data Glossary
- 3.5. Build a Data Dictionary
- 3.6. Discover Metadata from Sources
- 3.7. Profile the Data
- 3.8. Identify Relationships Among Data Sources
- 3.9. Capture Data Lineage
- 3.10. Organize the Catalog for Users
- Best Practices for Building a Data Catalog
- Conclusion
- References
Introduction
A data catalog serves as a centralized reference tool enabling various users to explore, understand, and utilize data sets effectively. It collects metadata from diverse sources to create a searchable inventory, enhancing metadata management across an enterprise.
Why Data Catalogs are Essential
The primary goal of a data catalog is to overcome the challenges posed by data sprawl across different stores, making it hard for users to find relevant data. By offering a unified view and built-in search capabilities, data catalogs ensure operational and analytics initiatives are more effective, supporting data-driven decision-making.
10 Steps to Building a Data Catalog
3.1. Document Metadata Management Value
Highlight the benefits of metadata management to data governance, emphasizing the improved data quality and operational effectiveness it brings.
3.2. Identify Data Stewardship Uses
Distinguish between data catalogs, business glossaries, and data dictionaries to utilize each for effective metadata management.
3.3. Design a Subject Area Model (SAM)
Develop a SAM based on business uses of data, indicating data's location beyond system constraints, crucial for the data catalog's structure.
3.4. Build a Data Glossary
Create an enterprise-wide business glossary in collaboration with business data stewards, providing a foundational knowledge base for data catalog content.
3.5. Build a Data Dictionary
Compile comprehensive descriptions and mappings of data entities to guide metadata integration into the data catalog.
3.6. Discover Metadata from Sources
Identify and record metadata sources across the organization's databases and repositories for inclusion in the data catalog.
3.7. Profile the Data
Generate informative data profiles to aid users in understanding catalog metadata, focusing on both technical and business metadata aspects.
3.8. Identify Relationships Among Data Sources
Uncover and document data relationships across different systems to facilitate comprehensive data understanding and usage.
3.9. Capture Data Lineage
Utilize ETL tools for data lineage documentation, tracking data origins and flows for error tracing and user understanding.
3.10. Organize the Catalog for Users
Design the data catalog with a user-centric approach, ensuring accessibility and ease of use for data consumers.
Best Practices for Building a Data Catalog
- Ensure data security and privacy through user permissions and sensitive data tagging.
- Foster collaboration with user interaction features like rating, commenting, and chatting.
- Develop user training programs for effective data catalog utilization.
- Establish a maintenance process to keep the catalog current with evolving data assets and business needs.
Conclusion
A well-planned and implemented data catalog is integral to modern data management, offering invaluable support to data governance, metadata management, and user empowerment. It paves the way for a more informed and efficient operational and analytical environment.
References
- Smith, Anne Marie, Ph.D. (2022). "How to build a data catalog: 10 key steps."
dbx_data-and-ai-governance-6sept2023.pdf. - databricks link
Jim DG, EDM keys
My career is managing data anywhere: goalie, janitor, whisperer, therapist
Role keys- alignment on IT architecture, plans, priorities, business partnerships internal & external
SDP - virtual teams > discovery > assessment > plan > design > test > train > rollout > support
From IT EA, EDM, Digital Transformation programs, create related program for EDG that is aligned, integrated w BUs, clients, vendor services to meet goals on EDM quality & services OKRs, DT, Compliance ( internal and external audits )
EDM - Enterprise Data Mgt - 4 R keys - data & services are: RIGHT, RELIABLE, RESPONSIVE, REACTIVE across the enterprise, clients and vendor services
DTCC DLT Architect part of EA services tower w other IT lines > governance support on DLT, data
EDM app services, infrastructure, support tools
Collibra for MDM or ?
Couchbase for distributed NoSQL w JDBC tools, reporting
Pulsar & Solace on global events
DLT w Firefly, Fabric
MySQL, Postgres, Oracle, Snowflake, MongoDB, Aws Aurora
Messaging > Kafka, MQ, ActiveMQ, Artemis ?
Tools - EA, SA, Archimate, Plantuml, MDM repo ( NOT commercial ones but maybe integrated w Oracle )
RajG on Data Governance
Deloitte on Data Governance Challenges
Top 15 #DataGovernance Challenges in Context of the #GenerativeAI and How to Mitigate Them
1. Bias and Fairness
2. Ethical Use and Misinformation
3. #DataPrivacy
4. #DataQuality Assurance
5. Explainability
6. Transparency
7. Verification and Authenticity
8. Regulatory Compliance
9. #DataSecurity Risks and Malicious Use
10. #IntellectualProperty and Copyrights Issues
11. End-User Accountability
12. #HumanAICollaboration Guidelines
13. Public Perception and Trust
14. Community and Stakeholder Engagement
15. Real Time Monitoring and Control
Addressing these data governance challenges requires multidisciplinary, holistic and proactive approach that involves collaboration between data scientists, ethicists, legal experts, and stakeholders.
How to Mitigate the Above Challenges:
1. Bias Audits
2. Diverse Training Data
3. #ExplainableAI
4. Transparency Guidelines
5. Anonymization and Privacy Preserving Techniques
6. Human Review of the Quality Assurance
7. Encryption and Access Controls
8. Regular Audits by Legal Expertise
9. Clear Ownership Policies
10. Intervention Protocols for Real Time Monitoring
Image Source: Deloitte
RajG Data Governance Concepts for Generative AI
Why is #datagovernance important in #GenerativeAI?
Data governance plays a pivotal role in fostering #innovation in the evolving AI landscape by ensuring responsible data practices, mitigating biases, and safeguarding privacy. A robust data governance strategy is the key to unlocking the full potential of your Generative AI use cases.
Other DG and EDM Concepts
Who should own Data Quality?
Ownership of data quality is a critical aspect of effective #datamanagement within an organization. While the specific team or role that owns data quality may vary depending on the organizational structure, size, and industry, there are several common approaches:
#DataGovernance Team:
Role: A dedicated Data Governance Team or Office is responsible for defining and enforcing data management policies, including data quality standards.
Responsibilities:
-Establishing data quality policies and procedures.
-Defining data quality metrics and benchmarks.
-Monitoring and enforcing data quality standards.
#DataStewardship Team:
Role: Data stewards are individuals or a team responsible for the management and oversight of specific sets of data.
Responsibilities:
-Ensuring data quality at the operational level.
-Resolving data quality issues and discrepancies.
-Collaborating with business units to improve data quality.
IT or Data Management Team:
Role: The IT or Data Management Team, including database administrators and data engineers, may be responsible for technical aspects of data quality.
Responsibilities:
-Implementing data quality tools and technologies.
-Monitoring and optimizing data quality processes.
-Collaborating with business units to understand data requirements.
#BusinessAnalysts or #DataAnalysts:
Role: Business analysts or data analysts who work closely with business units and understand data requirements can play a role in ensuring data quality.
Responsibilities:
-Profiling and analyzing data to identify quality issues.
-Collaborating with data stewards to address data quality concerns.
-Participating in the definition of data quality rules.
Quality Assurance (QA) Team:
Role: In organizations with a strong QA function, the QA team may be involved in ensuring data quality for systems and applications.
Responsibilities:
-Applying QA principles to data-related processes.
-Conducting data validation and testing.
-Collaborating with data owners and stewards.
Business Units and Data Owners:
Role: In a decentralized model, business units or data owners may have ownership of data quality for the data they generate and use.
Responsibilities:
-Defining and maintaining data quality requirements.
-Taking ownership of data quality improvement initiatives.
-Collaborating with data stewards and IT teams.
#ChiefDataOfficer (#CDO) or Chief Analytics Officer (CAO):
Role: The CDO or CAO may have a strategic role in setting the overall vision for data quality and ensuring alignment with business goals.
Responsibilities:
-Setting the strategic direction for data quality.
-Advocating for data quality best practices.
-Collaborating with executive leadership to prioritize data quality initiatives.
Image Source: Eckerson Group
Analytics COE Models for Enterprises
Sample Digital Transformation Project Roadmap Checkpoints
Potential Value Opportunities
Potential Challenges
Candidate Solutions
Step-by-step guide for Example
sample code block