With the boom in automated operation and maintenance, CMDB construction projects continue It emerges precisely because CMDB is the cornerstone of automated operation and maintenance. Regarding the concept, positioning, value, relationship with surroundings, and the pain points faced by the company, I will not elaborate here. In summary, CMDB is very basic, very important and very complicated. This article goes straight to the topic, mainly about how the CMDB should be built, including the construction goals, framework and guidelines.
The goal of CMDB construction
The essence of CMDB is to build data, and data must be accurate and usable to be valuable. Therefore, the core construction goal of CMDB is to establish a set of high-quality IT resource allocation master data.
CMDB construction framework
We use a layered framework to think and analyze the entire CMDB implementation process. As shown below: 
The frame is divided into vertical view Five layers include management domain, management collaboration, product coordination, production domain and tools. The management domain defines the management work related to CMDB. The production domain defines the execution work related to the CMDB. The underlying tools are the support for the execution. The management and product collaboration are the reversal of the connection between management and execution;
Horizontal is a time dimension, and the production of the entire CMDB can be clearly seen What is the process. Among them, model design, data sorting and initialization, and the establishment and implementation of the CMDB maintenance system are the key links of the process. Complete and accurate data are the prerequisites for subsequent data analysis and visualization, and external system integration consumption.
Guidelines for the construction of CMDB
1. Model design
Model design is an important part of CMDB construction, which is the skeleton of CMDB. It is an abstract modeling process and should be abstracted based on the status quo of enterprise IT resources and consumer demands. The specific process is as follows:
1) Environmental research: Investigate the organizational structure of the enterprise, the type and scale of IT resources, and IT processes to fully understand the current status of the enterprise’s IT resources.
2) Model design: Design according to the research situation. The content includes model objects, model attributes, and model relationships, which we call metadata. In the process of constructing metadata, we follow the following principles:
The application-centric principle, the detailed design around the application from top to bottom;
Minimize the design principle, that is, only realize the current operation The minimum model set required by the dimensional scene to avoid the granularity of modeling from losing control (emphasize that objects/attributes/relationships without consumption scenes should not be stored);
3) Model review: After the model is designed, a specification needs to be formed Document. In order to conduct final communication and review with leaders and operation and maintenance administrators in various fields, the purpose is to form a model design recognized by everyone and lay the foundation for subsequent CMDB implementation.
4) Model establishment: After the model design review is passed, we can model in the CMDB tool. Therefore, CMDB tools must be flexible enough to flexibly define model objects, attributes and relationships according to the different environments of the enterprise.
2. Data sorting and initialization
After the designed model is established in the CMDB tool, the next step is to sort out and initialize the data. Note that data combing should be performed on an application basis, that is, one or more application system-related configuration data are combed each time to enter, and the combing process method is precipitated at the same time, so that it can be extended to other applications. It is a 1 to N process. The advantages of combing by application are:
The overall combing method can be quickly precipitated, one application can be used, and other applications can be copied;
It can quickly verify whether the design of the model is reasonable;
It can produce results quickly, and is easy to display and report;
3. Establishment of maintenance system
The key to data quality assurance lies in the establishment of a data maintenance system, data consumption, automation and data verification and other technical means only Supporting role. The data maintenance system mainly includes two aspects:
a) Role definition
Specify a configuration manager role, and the configuration manager is responsible for the overall CMDB. Including the definition and maintenance of processes, models, and role permissions.
b) Process definition
The warehousing process of each configuration item;
The change process of each configuration item, each attribute value of the configuration item must be processed Know its source and process;
The process of deleting each configuration item.
The data maintenance system is the focus and difficulty of CMDB construction, and it is an important means of ensuring the quality of configuration data. Because this link involves the participation of all parties in the enterprise, it is more difficult to be promoted solely by the implementation personnel of Party B. It needs to be designed and implemented together with both parties.
Four. System Integration
System integration is the last link of CMDB, relying on CMDB tools to provide good open interfaces. Because the core value of CMDB is to provide configuration data for external system integration consumption, it must provide a rich and easy-to-use API interface to facilitate low-cost integration with third-party systems.
High-quality data is the ultimate goal of CMDB construction, and only by constructing its skeleton (model) and injecting the soul (maintenance and consumption) can CMDB truly come alive.