1. Data standard classification and formulation
(1) The concept of data standards
Data Standards: Data is a business asset. Identify and define business data, and develop and publish enterprise-wide data definitions and standards for critical data. Each data has a uniquely assigned owner who is responsible for defining the rules for use and protection of the data.
According to the big data standard system formulated by the big data standard working group of the National Information Technology Standardization Technical Committee, the big data standard system framework consists of seven categories of standards, namely basic standards, data standards, technical standards, platform and tool standards, Management standards, security and privacy standards and industry application standards.
The data standard development process includes the following activities: proposing requirements, developing business data standards (BDS), review, release, archiving, etc. Data standards are consistent agreements on the expression, format and definition of data, including unified definitions of data business attributes, technical attributes and management attributes; the purpose of data standards is to make the data used and exchanged within and outside the organization consistent and accurate.
Data standards are the basis of data quality management and provide a basis for judgment on data accuracy and consistency. Data standards are divided into basic data standards and indicator data standards. Basic data standards generally include master data standards, logical data model standards, physical data model standards, etc.; indicator data standards are generally divided into basic indicator standards and technical indicator standards. It is recommended that enterprises:
First, cover the data business dimension, technical dimension and management dimension in the data standard content framework. Among them, business dimensions include: name, business definition, decomposition dimensions, etc.; technical dimensions include: code format, value range, length, precision, etc.; management dimensions include: data responsible department, standard release and expiration date, etc.
Second, the data standard classification framework covers master data standards and indicator data standards. Among them, master data standards include: human resources master data standards, financial master data standards, procurement master data standards, etc.; indicator data standards include: enterprise strategy and control indicator data standards, enterprise investment and operation indicator data standards, and enterprise service and support indicators. Data standards, industrial sector operation indicator data standards.
(2) How to formulate data standards
Data standard management is to formulate data standards that meet the business needs of the enterprise and future development trends for the data within the enterprise, and implement and apply them correctly and in a timely manner in various departments and application systems; at the same time, according to the continuous changes and development of the business , update and maintain data standards and apply them to various departments and application systems to adapt to the latest business and ensure consistency with business goals. Data standards management is reflected in the formulation, review, implementation, and feedback of data standards.
Data standard management strategy: It is the basis of the data standard maintenance process, which mainly defines the responsibilities and rights of each functional department in the process of data standard formulation and implementation.
Data requirements definition: It is a dynamic and continuous process. A clear definition of business capabilities is the basis for defining data requirements.
Data standard definition: It reflects the understanding and business definition of data standards by multiple relevant business departments, and the participation of business departments is a key factor in the definition of data standards.
Data standard release and implementation: It is the basis for ensuring the application of data standards. A strong cross-departmental leadership is the basic guarantee for the implementation of data standards.
Data standard application feedback: It is a closed-loop process from demand to application. Smooth application feedback channels are a key link in making the data standards maintenance process enter a virtuous cycle.
Arbitration of data standard ambiguity: Since data standard is a cross-departmental standard for business applications, authoritative arbitration is the basis for data standard operation when there is ambiguity in the application of data standard.
Data standard revision: It is a continuous workflow of "requirement-definition-validation-release-application feedback" to ensure that data standards adapt to business needs in real time.
It is recommended that enterprises develop data standards through the following four major measures:
(1) Develop implementable implementation plans. Implementation plans should focus on implementability, and plans that cannot be implemented will eventually be abandoned. A plan that can be implemented must have an organizational structure and division of labor. What each person is responsible for, how to assess, and how to supervise must be included in the implementation plan.
(2) Correctly understand the purpose of data standard construction, which is to unify the data caliber within the organization, guide the construction of information systems, and improve data quality.
(3) Fully realize the difficulty of data standardization. It is necessary to obtain the support of the management decision-making level, improve the level of organizational management, and make preparations for long-term promotion.
(4) Establish a scientific and feasible data standard landing form. It is necessary to consider how to implement data standards into existing systems and big data platforms.
By formulating data standards and specifications that meet the actual situation of the enterprise, establish an enterprise-level metadata management system, and promote its implementation in various fields of the enterprise to support the construction of data bases and digital operations. Enterprises can refer to the data standard template (see Appendix 2 for details), and describe in detail the data standard content of each L5 level data (ie: attribute) from three perspectives: business, technology and management: First, business perspective: for example,
subject Domain, business object, logical data entity, data classification, business attribute, business definition and usage, business rule, synonym, etc.
Second, technical perspective: For example, physical tables, system fields, data types, data lengths, whether there is a list of allowed values, allowed values, data examples, etc.
Third, the management perspective: for example, the subject responsible for business rules, the subject responsible for data maintenance, the subject responsible for data monitoring, etc.
2. Data standard framework system
For enterprises, data standardization is the basic work of enterprise informatization and digitization. It not only improves data sharing, but also provides enterprises with unified data views, data specifications and code standards that meet industry standards.
Data standardization should follow the following principles:
1) Unified standard data definition: reduce the ambiguity of data definition.
2) Unified data model management: ensure the establishment of an overall, enterprise-level data model that can fully express a common and consistent data view across systems, applications, and business domains.
3) Unified data encoding rules: the unified standard management of data also includes unified encoding rules.
Data standards run through the entire data life cycle, and the framework of the data standard system is shown in the figure:
Operational technical specifications: including data collection specifications, data security specifications, data classification specifications, master data management specifications, data modeling specifications, metadata management specifications, etc. Operational technical specifications constrain the implementation and execution of data standardization based on technical operation and management requirements at the operation level.
Basic data standards: including business terms, business rules, naming conventions and coding standards. Creating data standards requires business domain knowledge to ensure there is a consistent understanding of the data standards within the organization from creation.
Object data standards: including data classification standards, master data standards, data source standards, transaction data standards, indicator data standards, label data standards and subject data standards. Object standards describe the classification of data objects, as well as the classification, definition, naming, description and management processes or specifications for each type of data object.
Architecture data standards: including data catalog, data model, data distribution and flow, data exchange, data services and metadata standards. Data standards need to be based on enterprise-level data architecture and define the acquisition and use of data from a logical level.
Application data standards: refer to the functional management and business management processes or functional requirements that need to be implemented when developing and deploying information applications.
Data standard guarantee mechanism: including data standardization organization and standardization system, responsibility recognition and performance, talent training, and data culture. The data standardization guarantee mechanism provides guarantee for data standardization work from the aspects of organization, system and working mechanism, responsibility recognition and performance.
Data standardization management tools: including data sharing, services, data standards, data catalogs, data models, indicator data, metadata, master data and other management tools. Data standardization work requires the support of technical tools. Operational technical specifications need to be implemented in data governance and data asset management related software, and a long-term mechanism for data standardization should be established from the aspects of management process and technology implementation.
All aspects of the data life cycle are inseparable from the support of data standards. The standards for each link of data are as follows:
1) When data sources generate data, they need to follow business terminology standards, reference data standards and master data standards.
2) When collecting data, you need to follow data element standards, metadata standards, data collection standards, etc.
3) Data classification standards, business rules, naming conventions, etc. need to be followed when storing data.
4) When processing data, it is necessary to follow data modeling specifications, data model specifications, ETL operation specifications, etc.
5) When data is applied, it is necessary to comply with the quantity specification, data distribution and data flow specification, data label specification, etc.
6) Data archiving needs to follow the data archiving specifications, etc.
7) When destroying data, it is necessary to follow the data destruction regulations, etc.
3. Data standards guarantee mechanism
Data standardization is a long-term and systematic work. In order to ensure the effective implementation of data standardization, it is necessary to establish a sound data protection mechanism. The enterprise data standardization guarantee mechanism includes management and control organization, system construction, responsibility recognition mechanism and performance evaluation, talent training, and data culture.
1. Management and control organization
data standardization needs to organize resources, build processes, conduct business, and implement implementation within the enterprise based on factors such as enterprise management requirements, management and control positioning, management models, and business characteristics. By building a professional team or personnel for data standardization, as well as professional Accurate division of labor, cross-domain collaboration and linkage, and form a smooth communication, consultation, and cooperation mechanism. The organizational design should meet the requirements of functional coverage, efficient collaboration, and leading innovation, and match the upgrade and development of the company's overall business, organization, and management models. The division of labor is clear, and each performs its duties. Strengthen the implementation and follow-up supervision of various tasks of data standardization .
2. System construction
Enterprises need to formulate relevant data standard regulations to provide rules, definitions and standard specifications for various tasks including data standard management when standardizing the management of data asset management systems.
3. Accountability mechanism and performance evaluation
Data standardization needs to formulate an effective accountability process based on principles, have a clear division of responsibilities, and determine data performance evaluation rules, performance evaluation steps and processes, etc.
4. Talent training
To carry out data standardization, enterprises need to establish a capability cultivation and construction system including a training system and a talent evaluation system, and clarify the knowledge and capability structure requirements of data talents and the professional talent training plan.
5. Data culture
should continue to promote the construction of data culture within the enterprise, strengthen the promotion of data culture concepts and cases, improve the data thinking of managers at all levels of the enterprise, build a data discourse environment, and integrate data into the operations of departments at all levels and business units models and ways of thinking, and create a good data culture atmosphere within the enterprise.
4. Implementation of data standardization
The implementation of data standardization involves many fields. Enterprises should carry out implementation work in an orderly manner according to certain steps and plans based on their own characteristics and specific conditions and on the basis of overall planning.
1. Implementation of data standard management
The implementation of data standard management should be implemented in batches and phases based on the basicity, criticality and urgency of the planned content. The implementation route of the standard management system can be divided into strengthening the basis of data standard service There are three stages: improving data standard overall management and control capabilities and forming data standard service capabilities.
2. Data classification implementation
Data classification is the basis for realizing data asset management. Reasonable data classification will provide better guarantee for data management and application. According to different management and application needs, data classification can be done from many different perspectives. and dimensions. Data classification should follow the following principles: scientificity, stability, uniqueness, consistency, and priority.
3. Data classification implementation
Data classification is defined according to certain principles and methods based on the sensitivity of the data and the impact on the victims after the data is tampered with, destroyed, leaked or illegally used. Enterprise data classification first starts from the perspective of security compliance and data protection requirements, and secondly considers data value and application needs, taking into account the comprehensive requirements of enterprise data management. Data classification principles include: compliance, execution, timeliness, autonomy, rationality, and objectivity.
4. Master data management implementation
Master data management implementation requires establishing effective organizational guarantees and institutional support in the initial stage, forming a data standardization and standardized management model, and then continuing to operate effectively.
5. Implementation of data indicator management
When implementing data indicator management, it is necessary to standardize the management of indicator data generated by basic data processing, build an indicator system, carry out scientific indicator classification, and establish a series of guarantee mechanisms.
6. Metadata management implementation
The implementation of enterprise metadata mainly includes four stages: requirement analysis, planning and design, tool implementation, and continuous operation and maintenance. The implementation of metadata can help q quickly find out the data property, establish a data asset catalog, quickly browse and retrieve data through a unified data map, maximize the value of data, and solve the problem of q data islands.
7. Data model management implementation
The data model undertakes business semantics upwards and realizes physical data downwards. It not only includes data dictionaries, but more importantly, includes business topics, business objects, data relationships, and data standard mapping. Therefore, data modeling based on data standards is an important starting point for the implementation of data standards. The success of data modeling can make the implementation of data standards more effective.
5. Data standardization management tools
The implementation and management of data standardization involves complex management processes and technical operations of organizational coordination. Therefore, it is necessary to rely on corresponding technical platforms and tools to support data standardization management. Data standardization management tools generally include data maps, master data management, data indicators, metadata management, data model tools, data exchange and service tools, data asset management, data development, data quality management, data security, and data cost management. These tools are described in detail in later chapters and will not be repeated here.