How to understand the CMDB's routines

CMDB success and failure, about the master of the CMDB repertoire of more and less, deep and shallow!

A few days ago in a project to summarize, write the CMDB configuration management norms, found there are many routines, this article is Lao Wang summed up the CMDB routines!

routine 1:cmdb name should be changed, called it Resource management

What is configuration? Indeed, many of the tools for configuration management are now being inherited, but I prefer the resource concept mentioned in puppet. Resources can be almost equivalent to the concept of objects, objects have attributes, resources have attributes, objects have methods, resources have actions, additional points, resources and states. Remember that all objects can be viewed as resources.

Why do I insist on renaming? From the reality of the situation, everyone said that the CMDB is the traditional discussion, automatic discovery, configuration items, configuration properties. In addition to the design and management of some forms, ignoring a real CMDB?

The real CMDB is to manage all of the internal IT resources!

routines 2:cmdb models have levels

In the model, the CMDB model is hierarchical, and I define it as a core model and an extension model.

• Core model. The core model is the relationship between the business, application, and host hosts, and other relationships can be recorded. With this model can basically run the subsequent automation and monitoring system, and second can effectively manage the host information on the public cloud.

the core model is by no means an infrastructure-level resource model!

• Extend the model. Extension model is to rely on the core model to expand, for example, based on the application needs to find some associated resource information, based on the host to find some of its associated dependent device information, such as cabinets, storage and switches, etc., constantly expanding the object model.

Adhere to the core model of the import, and gradually drive the surrounding supporting resources to improve, which is the application of the most important point of the CMDB.

the object relation of routine 3:cmdb should be simplified

From there, you can see that there are only three relationships in the CMDB model, and three relationships are as follows:

• Master-Slave relationship. This relationship is a strong parent-child relationship, the Lord does not exist, then it does not exist. Expressed as a schedule, belonging to an object-level relationship. It can be expressed by a schedule, which is expressed in an inline table in the Easyops platform.
• The dependency relationship. is an association between object attribute levels, for example, the server is placed on the cabinet, the cabinet is placed in a room, this is the object level relationship. Expressed by an object's property association.
• The connection relationship. The relationship between the host and the storage, the host, and the network device is the connection. This relationship is generated dynamically and is an instance-level relationship.

What is the difference between dependencies and connection relationships?

• Dependency is a one-to-many relationship, and the relationship is maintained by people, such as a lot of servers on the cabinet.
• A connection is a many-to-many relationship, and the relationship is caused by a "connection", such as a server connected to a switch. Can be achieved through automatic discovery, if the person to maintain, basically impossible.

Routine 4: Don't be too superstitious to automatically discover

Auto-discovery can reduce maintenance costs and cost in Chengdu, but I don't believe in this ability. An automatic discovery of the ability must have the process of manual intervention, such as the automatic detection of network card speed, abnormal time, must not enter the CMDB, and then automatically found in a scenario is not directly effective, for example, a machine in the process and port information need to do automatic monitoring, If the process and port information on the host is maintained by Autodiscover (in fact, it is simple), this requires the monitoring system to adapt to the situation of the process being suspended during the change period, which causes the machine's process information to be automatically discovered.

Think about the boundaries of automatic discovery and manual maintenance?

First, related to the resource state of the Change Division, in fact, should need human participation. For example, ip/server resources from the resource pool in and out of the process, changes in state will involve the monitoring policy changes automatically. From this dimension of state, it is easy to find artificial and automatic boundaries, but not the fill of state attributes.

Second, cross-group resource management needs process-driven, for example, firewall, IP address, server is a typical cross-group/Department management resources. The management and consumer of the resource require some process control. Of course, this place has improved, ah, if the management platform is perfect, can be through the platform to simplify the process of Kazakhstan. The management of DNS and load balancing resources is also a typical example.

Each line in the diagram is a CMDB management process, with the exception of initialize complete!

routine 5:cmdb to lead, team to understand

Leadership is important, leadership is involved and the team has a consistent understanding that the CMDB is hard to succeed. Many CMDB projects fail, not technically, but with people.

Speaking of consensus understanding, I think the CMDB concept, model, process, scenario, implementation method should be simple enough. The introduction of the CMDB is best to start with a scene, whether it's supporting the event or monitoring it.

Routine 6: The conceptual level of cloud computing is the level of the CMDB

In the CMDB system, there is a deep level, the conceptual level of cloud computing is the model level of the CMDB. When you build a model, you also need to build a layered capability that, after the ability is partitioned, will have an impact on ongoing deployment. Our practice testing out is the continuous deployment of standardized norms also need such a layered approach, the cross-border led to system management is not clear, monitoring is the same!

One thing I'm not thinking about is, is PAAs resources managed by the application of subordinate resource management or as an independent resource? Especially in the public cloud mode.

routine 7:cmdb is a snapshot of your IT resources and organization

This is a simple saying, the CMDB not only maps the IT resource model you manage, but also reflects your organization's management model. when an object cannot find owner, what is the problem you need to think about? When a process cannot be implemented, you also have to think about whether the management of the organization is complex or not well executed?

There are a lot of routines behind the CMDB, it is different from the automation system, it is more difficult to make a mis than to make a tool system, and to understand these routines is close to success!

How to understand the CMDB's routines