|0x00 Digitization is a kind of "conspiracy"
At the 2016 Yunqi Conference, Jack Ma mentioned five major trends in the future: "new retail, new manufacturing, new finance, new technology and new energy". The first one is "new retail".
So what is "new retail"? Teacher Ma said: “Offline and online retail will be deeply integrated, coupled with modern logistics, service providers use big data, cloud computing and other innovative technologies”; Xiaoyaozi said: “The entire process of commodity production, circulation, and service is due to the Internet, The widespread use of big data has become more efficient." In short, this "new" word is inseparable from data and the ability to apply data.
But e-commerce is already a highly digitized industry, so the new round of digitization will not start from a simple e-commerce scenario, but continue to promote the digitization of enterprises or government systems to improve office efficiency and allow organizations to pass Data to achieve the upgrade. Therefore, the concept of "cloud and nail integration" was brought up, because DingTalk + Alibaba Cloud has too much imagination.
Where does the digitization of enterprises or government systems begin? In DingTalk's words, it is "organization online, communication online, collaborative online, business online, and ecological online". Another level of abstraction is the digitization of "human, financial, and material affairs." For any company, "people's data", "financial" data, "things data" and "things data" must be the core assets of the company. Therefore, companies want to improve their own efficiency and embrace digitalization. In this era, these aspects must be accelerated online, and then the organization will be digitalized.
Is Ali so big, can it lead the "digitization" of so many companies? Obviously not. However, you can use your own influence to come up with this concept and define it as a future strategic direction, which will cause general anxiety. Then "cloud nail integration", waiting for these companies right in front of the track, I have everything needed for digital transformation, I am ready, do you want to get in the car?
For B-side customers, digitization is a kind of "scheming". By mobilizing everyone's strength, we will build digital infrastructure together, and then grow a new unicorn business. Although many people do not agree with this point of view, you have to get on the car when your peers get on the car.
However, "inadvertently inserting willows and willows into shadows", the matter of digitization, is very good news for practitioners in the data industry. Because digital assets are more valuable, our value is also greater. Today we talk about a concept that is easily overlooked: master data, unlike the dimensional modeling that has been talked about in the past, master data is also a set of theories, which can also provide a digital solution for companies that do not have typical Internet business scenarios. .
|0x01 What is the main data
"Master data" is not a very new concept. It has been proposed for some years, but it is a concept that fits the current trend. The “master data” corresponds to the digitization of “personal, financial and material affairs” and the entire technical system applied behind it. If the organization is to achieve digitalization, then the concept of master data cannot be bypassed.
In the "Master Data Management Practice White Paper (Version 1.0)" led by the China Academy of Information and Communications Technology in 2018, the concept of master data (MDM) is defined as follows: "refers to an organization that meets the needs of cross-departmental business collaboration and reflects the status and attributes of core business entities Basic information of the organization. Compared with transaction data, master data has more stable attributes, higher accuracy requirements, and unique identification."
Therefore, many data that data practitioners usually contact, such as traffic reports, user reports and even financial reports, are not master data, but company employees, customers, and physical assets are considered master data.
According to international practice, we summarize some basic principles of master data:
- Authority: As the most important data asset of an enterprise, master data is not designed for business systems, but should remain relatively independent. It serves but is superior to business systems that use master data.
- Overall: Master data exists beyond departments and processes to meet the needs of cross-departmental business collaboration. It can be considered as the "greatest common divisor" of the business process of all functional departments.
- Shareability: Master data is data that needs to be shared between two or more systems before it can be defined as master data. Therefore, master data must apply a technical architecture that can be compatible with various heterogeneous systems.
- Scalability: When designing master data, it is necessary to consider the possibility of future expansion. Therefore, the definition of master data items should comply with the principle of opening and closing, that is, open for extension and closed for modification. In principle, the defined master data items should not be modified again.
In an enterprise, information such as products, materials, customers, suppliers, employees, accounting subjects, etc., are all master data and require specialized systems to maintain and build.
Many people here have questions: Why do we want to do master data, and what is the use of master data? Master data is not just a standardized data asset, and once a problem occurs, it is usually not a minor problem. The problem here is not only about data leakage, but also about misuse.
For example, a customer complained to a bank, claiming that his privacy was leaked. However, the bank self-inspected, it seems that there is no problem: it is just that different systems store different mobile phone numbers of customers. When sending information to customers, the bank uses mobile phone numbers stored in other systems, and this mobile phone number is set by the user Became a "privacy" state.
Or, in traditional business, a headquarters often manages hundreds of branches in dozens of cities across the country. If the information among the various organizations is not uniform, and a customer stores many versions of information, then just proofreading these data It takes a lot of energy.
Students who have done import and export business know the complexity of import and export business, so it is especially difficult to do data warehouse for import and export business. However, no matter how difficult the model is, there will be a day to be able to figure it out, but if the main data system involved is not well done, and the warehousing, distribution and other business sectors are independent of domestic business, and multiple systems are developed, then I am afraid that the data is collated every day. It's horrible to torture people.
Therefore, master data maps the standardization capabilities of business systems, and is the most important communication content when communicating between "data center" and "business center".
|0x02 What is the use of master data
Many people will have new questions: What is the difference between master data and the "metadata" we usually mention?
In the concept of data warehouse, metadata consists of two parts, namely:
- Business metadata: Provide user-based information, such as metadata that records business description information of data items can help users use data.
- Technical metadata: support the system to manage and maintain data. For example, metadata about the storage method of data items can support the system to access the data in the most effective way.
Therefore, although the names of master data and metadata are similar, they describe different things. Metadata is "data describing data", such as data type, data definition, data relationship, etc., which is equivalent to Excel header information; and master data is used to represent the company's general data, not only including Excel headers, but also The content of Excel is included.
To give a more straightforward example, suppose we have a book in our hands. The metadata is the catalog of the book, and the main data is the text of the book. We can retrieve the required main data through the metadata.
Therefore, master data is similar to metadata. Only data that avoids fragmentation and supports the digital transformation of business through a standard data system is good master data.
Although each functional area will build its own core data, such as company data in the legal system and financial data in the advertising system, in order to ensure that the business is running, they usually make a set of themselves first, but from a global perspective, The company's data should be constructed by the HR system, and the financial data should be maintained by a special financial system, and they should ensure uniform standards.
So how to use the master data? In addition to providing standard data interfaces, it is more important to provide standard business data to the data center, and then the data center uses the standard data to accumulate standard business process data, so that historical information will not be due to the system The adjustment has lost statistical significance.
Therefore, when it comes to enterprise digitization and the messy data, you must first think of master data, and then let master data solve the following three problems:
- Data normative issues;
- Data duplication problem;
- Data consistency issues.
Master data is a way of solving "data islands", and is another data governance solution that is different from "dimensional modeling".
|0xFF How to manage master data
Many companies know that master data is important, but most companies do a poor job of master data, because master data also needs to be managed through a set of methods.
The first consideration for master data is modeling. More consideration for master data modeling is the display of tree-like hierarchical structure. Similar to the idea of domain model, the key processes and steps should be as follows:
- Create data objects;
- Create child objects of the data object;
- Create data item information for each data object or sub-object, including name, type, and other extended attributes.
Each data object and child object corresponds to a table in the database. The child object alone has no meaning and must exist together with the parent object. For each data item corresponding to the modeling, it is the data field information in the actual data table.
The second consideration of master data is the maintenance of data, including the way of data synchronization, how to perform version management after synchronization, and control the permissions of corresponding visitors, and so on. At the same time, after the data is maintained in a system, it is necessary to provide external access interfaces, including synchronization to the data center, and real-time interface for single data query.
In many cases, due to the complexity of the data, the master data may not always be maintained in the master data system. Depending on the actual situation, the corresponding maintenance method can also be defined. For example, professional data can be maintained by professional systems, such as financial systems and legal systems, but master data provides data service capabilities that business systems do not have, such as data query interface services.
Although the concept of master data is very old and the method is relatively simple, it is easy to overlook. For example, when doing data statistics within an organization, if the organization's reporting method can be arbitrarily specified, it is easy to figure out the structure of the graph or even the cyclic structure, which makes it impossible to continue the data report.
Therefore, no matter how small the concept is, it also deserves our attention. As mentioned in "Encouraging Learning": "If you don't accumulate steps, you can't reach a thousand miles; if you don't accumulate small currents, you can't become a river."