Shulou(Shulou.com)06/01 Report--

With the development of business and the continuous introduction of new IT technology, the scale of IT resources of application systems is getting larger and larger, and the complexity of IT architecture is also increasing. In this case, we need to use a variety of monitoring systems, different ways to perceive whether the business system is alive, whether it is good or not, and how the user experience is. Common monitoring system types include: basic environment monitoring, network monitoring, system monitoring, database monitoring, application monitoring, user experience monitoring and so on.

In this scenario, we in the previous article "how to improve monitoring problems, try to create a unified enterprise monitoring platform system!" "

As discussed in this article, we need a unified monitoring center to manage multiple alarm sources, process the alarm data, and provide consumable monitoring data. The overall architecture diagram is as follows:

▲ Unified Monitoring capability Architecture

There will be a question, what is the relationship between monitoring and the CMDB of the enterprise?

CMDB and Monitoring

Our understanding has the following two layers of relationship:

CMDB needs to provide necessary supporting data for the monitoring system to converge, three-dimensional and standardize alarm information.

CMDB also needs to open the channel to the monitoring system and automatically add the object to the monitoring system when a new object joins the CMDB. At the same time, when the configuration data changes, it can send the necessary alarm information through the monitoring system.

Let's talk about the first layer of relationship first. A monitoring system, such as zabbix, will send an alarm when a certain monitoring indicator of an object reaches a threshold: the XX indicator alarm and detailed information of the XX object. And the dependency relationship between monitoring items can be configured in zabbix to achieve alarm convergence and correlation.

But there is a problem, let's imagine a scenario: you are the DBA of a large 2C company. At 3 o'clock in the morning on a winter night, the wind is blowing from the northwest outside, suddenly the mobile phone rings loudly, and there is an alarm message that application system A database node 01 has a connection abnormal alarm. The content of the alarm message is limited, do you want to get up and turn on the computer for further processing?

It's complicated, isn't it? In fact, when an administrator receives this alarm message, he not only needs to know that there is something wrong with the database, but also wants to know more information. for example: what application system the database belongs to, what environment it is located in, whether it is a highly available cluster, who is in charge of the application, which application systems need to rely on the application system, whether the database has the latest configuration changes, and so on. In order to make further judgments and arrange the next steps: for example, in the early hours of the winter, whether to get up and turn on the computer. Then at this time, we need a system that can provide: application level topology, cluster information, module information, resource instances, relationships and other information, this system is CMDB.

▲ CMDB configuration Management

The integration and integration of the two

With CMDB, when the alarm occurs, we can let the alarm system go to CMDB to query the comprehensive configuration information related to this alarm object, in order to provide the most accurate, rich and standard alarm information. For example, in the previous scenario, if we know that database instance 01 belongs to the test environment of application An and has a high availability cluster, there is no user to access the database at present. Why does the administrator get up and turn on the computer?

On the other hand, if it is found that this database is the database of the production environment of system A, and because of the recent upgrade, it is currently in single point mode, and systems B and C need to rely on system A, then get up and deal with the failure as soon as possible. and inform B and C to start the appropriate plan mechanism to minimize the impact.

Here, CMDB needs to have the ability to provide data to the monitoring system, need to have the corresponding data query, read interface information, and can be easily integrated.

▲ BlueKing API Gateway & API Interface list of CMDB

On the other hand, CMDB also needs to actively synchronize its own data to the monitoring system. For example, do we launch a batch of new virtual machine nodes in a certain system, submit a work order, record the CMDB configuration information, and then manually configure it in the monitor? It's obviously not very reasonable, is it? At this time, CMDB is required to actively push new object information to the monitoring system, which issues agent, configures monitoring protocols, starts monitoring and so on, according to the existing monitoring template.

In addition, if CMDB scans and finds that the actual configuration information of a host is inconsistent with the information stored in the current CMDB database, should it also be alerted by the monitoring system and notified to the administrator for further processing?

So you see here, there is a close relationship between the monitoring system and CMDB. What is more fatal is that there is often more than one monitoring system in the enterprise. If each monitoring system has to be integrated with CMDB, it must be exhausted to death. There needs to be the concept of monitoring center and unified alarm management. We do not need each monitoring system to be directly integrated with CMDB. We only need to connect all monitoring systems to the unified alarm center module, which will dock with CMDB monitoring and share information. In this way, when each alarm is issued, it can be turned into a standardized, three-dimensional alarm based on the information in CMDB, rather than a flat alarm. Only such an alarm can really highlight the value.

Author: Zhao Haibing

-END-

Review of previous periods

[practical information] which of the four Oracle DBaaS deployment models are you using?

How to improve the monitoring problem, try to create a unified enterprise monitoring platform system!

Cloud computing | is the data secure on the cloud? What about DDoS***?

How to practice agile development of Scrum based on TAPD?

Jiawei attended the GOPS global operation and maintenance conference: in the peak era of operation and maintenance, the research and operation center may be the only choice.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.