With the transformation of the world economy from an industrial economy to a digital economy, data has gradually become a key factor of production, and enterprises have begun to manage data as a strategic asset. Data is generated from the business and carried in the IT system. Effective governance of the data requires the full participation of the business and the compliance of the IT system. This is a very complex system engineering. Practice has proved that only by building an enterprise-level comprehensive data governance system, clarifying the business management responsibilities of key data assets, relying on standardized system and process mechanisms, and building effective management platforms and tools, can the value of data be truly realized.
1. Data Governance Architecture
The process of building a data governance system is the process of creating a "virtuous circle closed-loop data governance management system" with data applications as the core. After each IT system acquires all kinds of data generated by business activities, through systematic data governance and management, it continuously mines and realizes the value of data, expands and deepens data application scenarios, and guides business decisions. Problems, through the continuous revision of the data governance and management process, promote the overall upgrade of the business system, truly optimize the business process management mechanism and norms, and finally build a closed-loop management mechanism of data "acquisition→management→realization→discovery→response→correction". With the core of data application and the support of data governance platform tools, under the guarantee of data governance organization/system, we will continue to promote the realization of data standardization and business standardization through data governance means, and realize the effective linkage of business, technology, management and platform.
In the comprehensive data governance system, the core modules of data governance include data governance planning, data governance functions, and data governance platform tools. Data governance planning refers to the data governance system and planning, data governance organization and responsibilities, data governance systems and processes, and is The core module of governance and standardized management; data governance functions include eight functions: data standard management, data quality management, data architecture and model management, data development, metadata management, master data management, data lifecycle management, and data security management. , Enterprises usually merge management; data governance platform tools include data development platform, data asset management platform, data quality management platform, and data service platform. Usually, data governance platform tools are not completely consistent based on the stage functions of data governance. In practice, platform tools Usually combine multi-faceted functions, rather than single-platform functions.
The three modules are the driving force for each other. The data governance planning guides the full play of data governance functions. The various functions of data governance are assisted in management through data governance platform tools. The data governance platform tools support the implementation and optimization of data governance planning. Solidified on the data governance platform, the data governance platform assists the management of various functions of data governance, continuously implements and improves data governance planning through various functions of data governance, realizes the digital transformation of the organization, and solidifies the management mechanism and process system.
2. Data governance model
The data governance model refers to the data governance strategy that enterprises use to carry out data governance activities based on different data governance objectives and the status quo of enterprise organization, systems, and data applications. According to the 8-year practice of Kangaroo Cloud, the data governance model usually includes three basic models.
Mode 1: Bottom-up, focus on data architecture, and carry out data governance.
This model focuses on data architecture, and manages data layer by layer up to the data application layer. This mode starts from the underlying data, and based on the existing data foundation, inventory, construction, governance, and application are carried out layer by layer. It has high requirements for the overall data thinking and data governance level of the enterprise, and is usually suitable for heavy data and light business applications. Technology-based enterprises, or government agencies, or enterprises with many new and self-developed systems.
Mode 2: From top to bottom, focus on clear data application and carry out data governance.
This mode is a single-point application mode, and data governance is usually carried out with the existing application requirements as the core. Focus on the data application and data governance needs of various business fields, and organize and promote data governance work on demand under the premise of demand, resources and driving force. Data governance can only be done well with the in-depth participation of the business department, and only the governance that is tailored to the needs of the business can be recognized and supported by the business department.
This mode usually carries out data governance around the needs of data applications, such as upgrading the architecture, changing platforms, etc. when data application migration is involved, or when focusing on supervision, reporting, etc. to clarify data applications, data governance is carried out around data applications.
This model is usually suitable for enterprises with strong data applications and strong business departments, but weak overall data awareness. The data governance entry of this model is relatively simple. Practice has proved that most enterprises will use this model in the early stage of digital transformation, and slowly explore the data governance path of the enterprise. This model will help align the cognition of data departments and business departments. , improve the overall data awareness of the enterprise, and provide the cornerstone for the development of future data governance.
Mode 3: Large-scale planning mode, starting from data application planning, governing the current situation, planning for the future, and carrying out data governance based on the future of data assets.
This model requires enterprises to comprehensively sort out the current pain points of the business and the future vision of the business, take stock of the current situation, plan for the future, plan and analyze application scenarios based on the current and future needs of the business, and comprehensively sort out the current status of the data within the scope of the application scenario blueprint planning. Plan the future of data, and formulate a comprehensive strategy for the data requirements in the blueprint planning. Which new systems and new data sources will be purchased? Which ones need to upgrade existing data systems, refine and standardize existing data? Which data requirements are more feasible to implement? Formulate a comprehensive planning system, divide priorities, and implement comprehensive data governance in a rhythmic and step-by-step manner. This mode is usually a strategic project of an enterprise, which is promoted by high-level executives and has high requirements for data and business synergy. The whole process involves system transformation and upgrading, business process optimization and reengineering, and is a process of comprehensive upgrading of the enterprise.
3. Data governance implementation path
The first stage: the start-up stage, the digitalization stage of business operations.
This stage is mainly to sort out the status quo facing the enterprise, respond to pain points, and explore business scenarios. Enterprises gradually begin to transform from informatization to digitalization. At this stage, enterprises will re-examine the original data governance strategy, reconstruct the data governance strategy and implementation path, gradually begin to build a data governance framework, a data governance system framework, and upgrade the original data processing , Application mode, build a big data platform, build the basic capabilities of big data collection, collection, storage, calculation, and service, gradually integrate data from various systems, break data islands, accumulate data assets, and explore business scenarios.
The second stage: the stage of in-depth expansion, and the stage of normalization of data empowerment.
At this stage, data application becomes the focus, and enterprises begin to dig deep into the value of data to increase the coverage of data applications. The scope of data application, from the realization of core KPI indicators, gradually covers all core businesses, builds a complete analysis framework and insight system, and continuously improves the quality of business decision-making. The big data platform continues to develop the ability of big data processing. The enterprise incorporates more and wider data content, continuously expands the breadth and depth of data application, and initially forms the enterprise's data asset map. The data standard system is gradually established, and the efficiency of data application is greatly improved. Improvement, initially completed the transformation from "empiricalism" to "dataism", and data decision-making has become the main decision-making method of enterprises.
At this stage, the enterprise began to fully establish a data management authority system, improve the data governance mechanism, optimize the data governance process and system, upgrade from the original "extensive" management to "fine" management, and continuously improve data quality. Enterprise data management Capability upgrades, and gradually realize intelligent management through data quality platforms, data asset platforms, and data governance platform tools, and comprehensively improve the enterprise's data thinking cognition.
The third stage: the stage of intelligent application, and the stage of intelligent operation decision-making.
At this stage, enterprises have realized the integration of insights and strategies, smart scene applications have become the norm, fully completed digital transformation, explored digital business, and opened a new chapter. This stage is dominated by intelligent applications, and AI empowerment has become the norm. Enterprises continue to mine the value of data, stimulate innovation, and begin to provide accurate data dependence for strategic analysis. At this stage, some enterprises even rely on their original business models. Inspire new business models.
At the level of data management, the construction of the data governance system is gradually advanced to the optimization of the data governance system, the mechanism and process are improved, and the data management responsibilities are further refined; at the level of data assets, the construction of global data assets is completed, a strong data model system is built, and the enterprise data management system is completed. Standard construction, constantly improving the data asset system; at the platform tool level, the capabilities of big data platforms are gradually shifting to algorithmic capabilities, the development of intelligent recommendation algorithm models has become a normal demand, and the data governance platform is gradually improving its functions to assist enterprises in intelligent data quality and data standards , data assets, master data and other modules, the enterprise has truly entered the stage of intelligent operation decision-making.
4. Data governance practice of large-scale digital twin project
A high-speed rail hub station has a total construction area of 1.22 million square meters, an annual passenger flow of 137 million, and an east-west span of 820 meters. It is one of the largest railway hub stations in Asia, and one of the most important modern comprehensive transportation hubs in China. The hub station involves more than 50 management and cooperation units across fields, regions, departments, levels, businesses, and affiliations, and integrates multiple transportation transfer methods. In daily operation, faced with a series of problems such as complex station structure, difficult management and control, difficulty in responding to sudden large passenger flow, prevention and control of epidemics with dense crowds, emergency rescue in extreme weather, and coordination and linkage of public security management and control, in order to better Transferring management experience from offline to online requires a large amount of data to support it.
According to the project construction requirements, a high-speed railway hub station digital governance laboratory and special work class were established to build a "four-in-one" driving mechanism involving the participation of railways, localities, institutions, and society, and improve the "four horizontal and four vertical" security systems. Form a pattern in which the main drive of the operating unit, the deep collaboration of the research institution, and the sharing of all parties in society.
According to the V-shaped model theory, business dismantling, data collection, and function integration are carried out, and a data base with "IoT (IoT) + GIS (Geographic Information System) + BIM (Building Information Model) + public data" as the core is established , Through data analysis, interactive feedback, and algorithm model support, establish a 3D digital twin system to promote visual management of traffic operation, resource allocation, and force deployment, and realize refined and collaborative governance of large-scale comprehensive transportation hubs.
One is to clarify the application tasks. Focusing on the needs, five first-level tasks, including passenger flow forecast and early warning, smart epidemic prevention, meteorological disaster prevention, fire safety, and public security prevention and control, are clarified, and they are detailed to the smallest granularity one by one. For example, meteorological disaster prevention includes 6 second-level tasks such as rain, snow, hail, freezing, strong wind, and high temperature, as well as 18 third-level tasks and 63 fourth-level tasks such as traffic guidance and anti-skid disposal.
The second is to establish an indicator system. Around the five first-level tasks, quantifiable index systems were established respectively, and the pressure index, traffic index, retention index, and early warning index were generated through comprehensive calculations to directly reflect the overall situation of operation.
The third is to open up and collect data. The China Unicom system platform collects more than 200 items of data through various methods such as calling, accessing, and collecting to form a safety prevention and control data system for a high-speed rail hub station.
The fourth is to realize comprehensive integration. Focus on creating "one picture, one database, one network + N sub-scenes".
One picture, that is, the digital twin panoramic operation diagram, uses 3D modeling, simulation and other technologies to visually manage traffic operation, resource allocation, force deployment, etc. in the form of panoramic views and index charts, so as to realize the effective control of the overall situation and precise focus synergy.
One database, the thematic database, is an integrated intelligent public data platform in China Unicom District, which stores and collects basic data, directory data, event data, etc., and generates governance data through edge, cloud, and supercomputing.
One network, that is, the intelligent management network of a high-speed railway hub station, relies on the department management terminal and mobile execution terminal, through information integration, business integration, and execution integration, to establish a complete work chain of item discovery, reporting, circulation, disposal, and feedback. The main body collaboration link realizes the "integrated coordination" of cross-department, cross-system, cross-region, and cross-level matter handling.
N sub-scenes: According to the principle of urgent use first and maturity first, four sub-scenes of passenger flow forecast and early warning, meteorological disaster prevention, fire safety, and public security prevention and control will be created in advance.
