Maturity of master data management (MDM)

MDM Full write Master Data Management, Translation is the main data management or metadata management .

What is? MDM？

Enterprise master data is the data used to describe the core business entities of an enterprise , Such as customers 、 partners 、 staff 、 product 、 Bill of materials 、 Accounts, etc ; It is of high business value 、 Data that can be reused across business units within the enterprise , And it exists in many heterogeneous application systems .

The concept of master data and master data management

Enterprise master data can include many aspects , In addition to the common customer master data , Customers in different industries may also have other types of master data , for example ： For customers in the telecommunications industry , Various services provided by telecom operators can form their product master data ; For airline customers , routes 、 Flight is a kind of its business owner data . For different business departments of an enterprise , Its master data is also different , For example, the marketing department closes Customer information , The product R & D department cares about the product number 、 Product classification and other product information , The personnel department cares about the staff organization , Department level relationship and other information .

As shown in the figure , The content and scope of enterprise data management usually include transaction data 、 Master data and metadata .

Trading data
Used to record business events , Such as customer's order , Complaint record , Customer service application, etc , It is often used to describe the behavior of the business system at a certain point in time .

Master data
Master data defines the core business objects of the enterprise , Such as customers 、 product 、 Address, etc , Different from transaction flow information , Once the master data is recorded in the database , It needs to be maintained frequently , So as to ensure its timeliness and accuracy ; Master data also includes relational data , Used to describe the relationship between master data , Such as the relationship between customers and products 、 The relationship between products and regions 、 Customer to customer relationship 、 Relationship between products, etc .

Metadata
That is, data about data , Used to describe data types 、 Data definition 、 constraint 、 Data relation 、 Information such as the system where the data is located .

Master data management refers to a set of specifications for generating and maintaining enterprise master data 、 Technology and solutions , To ensure the integrity of the master data 、 Consistency and accuracy . Typical applications of master data management include customer data management and product data management .

Information flow of master data management

Generally speaking , The master data management system is from IT From the perspective of construction, it will be a relatively complex system , It is often associated with enterprise data warehouses / The decision support system and all business systems in the enterprise are related , Technical implementation will also involve ETL、EAI、EII And so on , Pictured 2 Shown , A typical master data management information flow is ：

A business system triggers changes to the enterprise master data ;
The master data management system will be integrated 、 Accurate master data is distributed to all relevant application systems ;
Master data management system provides accurate data source for decision support and data warehouse system .
Therefore, for the construction of master data management system , We should consider the overall platform framework and technical implementation from the initial stage of construction .

MDM The meaning of

Integrate 、 share 、 Data quality 、 Data governance is the four elements of master data management , What master data management needs to do is to integrate the core from multiple business systems of the enterprise 、 The data that needs to be shared most （ Master data ）, Centralized data cleaning and enrichment , And unified in the way of service 、 complete 、 accurate 、 Authoritative master data is distributed to operational and analytical applications that need to use these data throughout the enterprise , Including all business systems 、 Business process and decision support system .

Master data management enables enterprises to centrally manage data , Ensure the consistency of master data among decentralized systems , Improve data compliance 、 Quickly deploy new applications 、 Fully understand the customer 、 Speed up the introduction of new products . from IT From the perspective of construction , Master data management can be enhanced IT Structural flexibility , Build a data management foundation and corresponding specifications covering the whole enterprise , And more flexibly adapt to the changes of enterprise business needs .

Take customer master data as an example , Customer master data is a common problem faced by enterprise customers , In most businesses , Customer information is usually scattered in CRM And other business systems , In each business system, there are only fragments of customer information , Incomplete customer information , But it lacks enterprise level integrity 、 Unified single customer view , As a result, enterprises cannot fully understand customers , Unable to coordinate and unify market behavior , Lead to a decline in customer satisfaction , A decrease in market share . therefore , The purpose of establishing customer master data system is ：

Integrate and store the information of customers and potential customers of all business systems and channels ： On the one hand, extract customer information from relevant systems , And complete the cleaning and integration of customer information , Establish an enterprise level unified view of customers ; On the other hand , The customer master data management system synchronizes the unified customer information formed to other systems in the form of broadcasting , So as to ensure the consistency of customer information .
Provide online transaction support for relevant application systems , Provide the only access point to customer information , Provide timely and comprehensive customer information for all application systems ; In the service of OCRM System , Make full use of the value of data , Provide more value-added services at all customer touchpoints .
Realization SOA Architecture of ： Before establishing the customer master data system , Data is locked in every application system and process , After establishing the master data management system , Data is released from the application system , And it is processed into a set of reusable services , Called by various application systems .

Master data management (MDM) The maturity of

Depending on the complexity of the implementation of master data management , In general, master data management can be divided into five levels , From low to high reflects master data management (MDM) Different maturity .

Level 0 ： No master data management is implemented (MDM)

stay Level 0 Under the circumstances , It means that there is no data sharing between the various applications of the enterprise , There is no data definition element in the whole enterprise .

such as , A company sells a lot of products , The production and sale of these products are handled by several independent systems , Each system processes product data independently and has its own independent product list , Product data is not shared between systems . stay Level 0, Each individual application is responsible for managing and maintaining its own critical data ( For example, the product list 、 Customer information, etc ), This information is not shared between systems , The data is disconnected .

Level 1 ： Provide list

Whether the company is big or small , List management is a commonly used method . Within the company , Maintain a logical or physical list manually . When various heterogeneous systems and users need some data Hou , You can get the list . For the maintenance of this list , Including data addition 、 Delete 、 Update and conflict handling , It is handled by the staff of various departments through a series of discussions and meetings .

Business rules are used to reflect the consistency of value , When business rules change or similar situations occur , This highly manual management process is prone to errors . Because list management is managed manually , The quality of its list maintenance depends on who participates in the change management process , Once someone is absent , It will affect the maintenance of the list .

Level1 Than Level 0 The difference is , Although each department maintains its own key data independently , However, a loose master data list will be maintained through list management , Be able to provide other departments with the data they need . stay Level 1 in , Data change decisions and data change operations are all decided by people , therefore , Only after people complete the data change decision can they change the data .

In practice , Although the data change process has strict regulations , But due to the lack of centralized 、 Rule based data management , When the amount of data is large , The cost of data maintenance will become very high , And it's going to be very inefficient . When master data , Such as customer information 、 When the quantity of product catalog information is relatively small , list The way of management is feasible , But when the product catalog or customer list explodes , The change process of list management will become difficult .

Level 1 Rely on human collaboration . Implementing a customer or product list on an enterprise scale is like maintaining the relationship between people in different departments . If there is a hierarchy or grouping of customers or products , The list will be hard to provide , And usually in Level 1 Because it is too complex to be managed .

Level 2 ： Equal access ( By way of interface , Each system is directly interconnected with the master data host )

Level 2 And Level 1 comparison , The introduction of ( Automatically ) management . By establishing data standards , Define the access and sharing of detailed data stored in the central knowledge base , It provides strict support for sharing usage data among various systems . A central knowledge base is often referred to as the primary data host . This knowledge base can be a database or an application system , Support data access and sharing through online means .Level 2 Introduced “ Equal access ”, In other words, an application can call another application to update or refresh the required data .

At this stage , Rule management 、 Data quality and change management must be customized across the enterprise as additional functions . stay Level 2, Data changes are done automatically — Through standard processes implemented by specific technologies , Allow multiple application systems to modify data .Level 2 It can support different application uses and change orders 、 Shared data knowledge base .

Level 2 Each equivalent application needs to understand the basic business rules in order to access the main list 、 Interact with the main list . therefore , Every equivalent application must be properly created 、 increase 、 Update and delete data . Authorized applications are responsible for adhering to data management principles and constraints .

Level 3 ： Centralized bus processing

And Level 2 comparison ,Level 3 Breaking the organizational boundaries of individual applications , Use data standards acceptable to all systems to uniformly establish and maintain master data (Level 2 The data stored on the master data host of is still stored separately according to each system , There is no real integration ). Centralized processing means for MDM Build a general 、 Target based platform .

Most companies find MDM Are challenging their existing IT framework ： They have too many independent platforms for processing master data .MDM Level 3 Centralized data access 、 Control the use of data across different applications and systems . This greatly reduces the complexity of application data access , It greatly simplifies the management of data rules , send MDM It has more functions and features than a decentralized environment . Enterprise master data faces the challenge of consistency . Data exists in different places , The meaning of data is also different , The rules of data vary from system to system . focus MDM Handle - Through a public platform as a bus (HUB)- Explain a consensus , Integrate subject domain data from multiple systems , It means using centralized 、 Standardized methods to transform heterogeneous operational data , No matter what it looks like in the source system , Will be integrated .

stay Level 3, The company adopts a centralized management mode for the subject domain content . This means the application system , As a consumer or using master data , There is a consensus that data is an image of the content of subject data , Breaking the organizational boundaries of individual applications . stay Level 3, A company can let any two systems share data and speak each other's language .Level 3 It also reduces the complexity of equivalent access ." consumption " Applications no longer need to support system positioning and operation logic .

Any distributed details related to the source system data will be MDM Bus centralized processing . stay Level 3 Automatic data standards mean ： Establish the target data value representation and provide accurate master data value capture through the necessary steps . In all categories, from Level 3 Start supporting consistent enterprise data views for the first time . Data quality rules are here for data cleaning and error correction .

Level 4 ： Business rules and policy support

Once data is integrated from multiple data sources , The topic domain view goes beyond individual applications and is represented as an enterprise view , You will get a single version of the facts . When a single version of the fact is available , The inevitable response from business executives and executives is often ：“ Prove it ”.Level 4 It can ensure that the master data reflects the business rules and processes of a company , And confirm its correctness .

Level 4 Support rules by importing master data , Also on MDM Check the integrity of bus and other external systems . Because most companies are relatively complex , Rules and policies that affect business data access and operation Relatively complex . It is impractical to assume that any single system can contain and manage various types of rules related to master reference data . therefore , If one MDM The bus is really intended to provide enterprise wide data accuracy , Support for workflow and process integration is essential .

MDM The system must not only support rule-based integration , We should also be able to integrate external workflows . These rules may include interacting with a clinical system through a bus or waiting for another system or person ( People who have the authority to make changes ) The examination and approval . Through one MDM Bus , Rule definitions can be more than just logical , It can also rely on inputs from other systems .

Of course , Coordinating and auditing data means that other systems can be rolled back ( Or business process ) To ensure that data changes are strictly approved , In this way, errors can be found and transactions can be rolled back when needed .Level 4 Propose support for rule and policy extensibility . It is important to support any business oriented rule set in a flexible and sustainable way through the bus .

such as , If a store manager updates the price of a product , The bus system needs to be able to communicate with a trusted system ( such as , Commodity management system ) Negotiate to make the rules effective . Detailed rules will support changes in product prices in another system — The bus needs to be able to understand the authorization system or method that can process and approve changes . These rules may involve complexity or privacy restrictions , They are forbidden to exist directly on the bus .

stay Level 4, An enterprise can support a set of steps or tasks , In a special creation 、 Read 、 These steps or tasks must be followed before updating and deleting tasks are allowed . Workflow automation is often used to support what happens on the bus Authorization of a piece or activity . But change management is much more than workflow ： It can include logic based processes and human based decisions . The existence of change management can support dynamic business , Allow changes .

Level 4 Support centralized rule management , But the rules themselves and related processing can be separated . let me put it another way ,MDM The bus needs to ensure that the rules are applied centrally , Even if this rule lives outside the bus .

Level 5 ： Enterprise data sets

stay Level 5 , The bus and related master data are integrated into independent applications . There is no obvious separation between master data and application data . They are one . When master data record details are modified , All application related data elements will be updated . This means that all consumer applications and source systems access the same data instances .

This is essentially a closed loop MDM： All application systems are integrated through master data under unified management . At this level , All appear to be the same version of the truth in the system . Operate the application system and MDM The content is synchronized , So when a change occurs , The operating application system will be updated . In those familiar MDM Architecture style , Persistent bus architecture , When a bus updates all the operating application systems, it will reflect this change , Form a direct operational view of change . In the registration environment , When the data is updated , Pass... Through the bus Web Service connection related system application transaction update .

therefore ,Level 5 Provide an integrated , Synchronous architecture , When an authorized system updates a data value , All systems in the company will reflect this change . After the system updates the data value, do not single select the update of the corresponding value in other systems ：MDM Will make this update transparent .

A company is completing MDM Level 5 After that, all their applications will be connected — Both operational and analytical — All access to master data is transparent .Level 5 Is to take the concept of data as a service To achieve .Level 5 Ensure a consistent master data subject domain enterprise image . Definition “ Customer ” It is actually the same thing with other applications to accept changes in business rules of customer master data .Level 5 Removed the last obstacle of master data ： Unified data definition 、 Authorized use and change dissemination .