当前位置：网站首页>Multi dimensional monitoring: the data base of intelligent monitoring

Multi dimensional monitoring: the data base of intelligent monitoring

2022-06-28 13:39:00 【Tencent blue whale assistant】

Preface

Take component monitoring as an example , Introduce the roadmap for monitoring products

The function of operation and maintenance monitoring system is self-evident , Throughout the operation and maintenance 5 Functions ： Release 、 change 、 Fault handling 、 Experience Optimization 、 Daily needs , Ensure the service availability of the above functions .

From the characteristics of big data ( Large amount of data 、 multidimensional 、 completeness )1 Look at , The construction of operation and maintenance monitoring system can be divided into 2 Stages ： Multidimensional monitoring （ Accumulate data ） and Intelligent monitoring ( Using data ）, Through multi-dimensional monitoring, the fault can be seen 、 Yes , Intelligent monitoring detects risks in advance 、 Find out the cause of the fault .

Insert picture description here

Component monitoring is the first step of a multi-dimensional monitoring system 3 layer , Mainly for Common open source components 、 Monitor the performance indicators of middleware , such as Nginx The performance indicators are Active Connections（ Current number of client connections ）、Waiting（ Number of connections waiting ） etc. ,Oracle The performance indicators are SQL Hard resolution rate 、 Table space usage, etc .

By collecting the key performance indicators of components , Learn the health status of components in real time , Find problems ahead of time , Instead of just monitoring whether a process or port is alive （ When the process or port is normal , Does not mean that services can be provided ）.

This paper takes the construction component monitoring as an example , from The composition of multi-dimensional monitoring 、 Monitoring the product to solve 3 A question 、 Technology selection of component monitoring 、 Cloud distribution collector configuration 、 The openness of the community To introduce the monitoring product design roadmap .

1. The composition of multi-dimensional monitoring

From the perspective of user access to the link , The dimensions of monitoring indicators are divided into User level 、 application layer 、 Component layer 、 Host layer 、 The network layer .

The composition of multi-dimensional monitoring

User level , Simulate the user's access behavior through service dial-up testing , You don't have to wait for users to complain ; application layer , Trace the call status of the application through the call chain ; The other three layers are easy to understand and will not be introduced .

Through this 5 layer + Other key indicators （ Like a journal 、 Business KPI Curves, etc ）, Build multi-dimensional monitoring capability of monitoring system , Provide data support for the second stage of intelligent monitoring .

2. Monitoring the product to solve 3 A question

In addition to obtaining key performance indicators , Monitoring products still need to be solved 3 A question , Failure correlation analysis can be carried out for fallback , The intelligent scenario of operation and maintenance can be built .

2.1 Yes IT Autonomous Control of the system

because Yes IT Lack of autonomous control ability of the system ," Replacing IT System " and " Trend replacement IT On the way to the system , Is part of 、 Large enterprises in " Internet +" Actively embracing the current situation of the Internet under the tide .

In view of this situation , Some industries have made it clear that 2, We must pay more attention to IT The ability of the system to control itself .

therefore , Product design , It should be considered that users of the monitoring system can participate in the development or partial development of the monitoring system .

2.2 Refuse to build another chimney

The shaft structure is estimated to be built by most enterprises IT The state of the system , There is no correlation between each system , Each purchase of a system is equivalent to building an information island , Extremely low added value .

Insert picture description here

If you want to realize fallback, you can perform fault correlation analysis , The intelligent scenario of operation and maintenance can be built , Can be based on PaaS On the operation and maintenance platform 4, adopt iPaaS Get through all the inside of the enterprise IT Operating system .

2.3 There are many components , It's not very realistic to be completely self-study

There are a wide variety of components used in the industry , From database 、 Storage 、HTTP Service to message queue, etc 100+, It's certainly unrealistic to make a complete self-study .

A good way is to study the core by yourself 、 Components with poor industry support , The rest rely on the accumulated capacity of the industry for many years , Make fewer wheels , Save electricity for the society .

3. Technology selection of component monitoring

stay 2.3 Self research is mentioned in + The first 3 The idea of open source collector , Here is the open source collector Prometheus Exporter For example .

Insert picture description here

Prometheus Exporter Our community is very active 5, Support 100+ Common open source components , Some large factories even specially write corresponding Prometheus Exporter, such as Oracle Compiling Weblogic Exporter,IBM Compiling IBM MQ exporter,k8s、etcd Even built-in based on Exporter canonical metrics.

According to this scheme , Just do one Protocol conversion You can stock in indicators

Connect with industry collectors Simple sequence diagram

4. Experience Optimization ： Cloud distribution collector configuration

After solving the basic requirements , You need to optimize your experience right away .

Send the collector or configuration to the monitored host , Generally, you need to manually deploy or use third-party tools （ Such as Ansible）.

Switch multiple systems to accomplish one thing , The experience is very bad .

There is an optimization scheme , adopt iPaaS Use the file distribution and command execution capabilities of the control platform layer 4, Let users complete the configuration process in one page , Improve efficiency .

Blue whale architecture diagram

5. The openness of the community

After meeting the basic functions and optimizing the product experience , Next, consider Product scalability .

First, it solves the convenience of users' one click Import of self-developed components , Next, provide a communication platform for community users to share freely .

While gaining the open source capability of the community , It also needs to feed the community .

6. ending

The multi-dimensional monitoring that belongs to the basic monitoring scope is relative to the intelligent monitoring , Not very bright , but It is the data base of intelligent monitoring , There is no data provided by multi-dimensional monitoring , Failure prediction cannot be realized 、 Intelligent monitoring scenarios such as fault root cause analysis .

When traditional enterprises or Internet enterprises embrace the change of the Internet , Need to think calmly , Follow the roadmap step by step .

7. reference

1 Wu Jun . The age of intelligence ： Big data and intelligent revolution redefine the future M. Beijing ： Citic publishing group ,2016-8.

2 People's Bank of China . Information technology in China's financial industry “ Much starker choices-and graver consequences-in ” development planning EB/OL. 2017.06

3 China Banking Regulatory Commission . China's banking information technology “ Much starker choices-and graver consequences-in ” Regulatory guidance on Development Planning （ Solicitation draft ）EB/OL. 2016.07.15

4 China Communications Standardization Association . Cloud computing operation and maintenance platform reference framework and technical requirements EB/OL. 2017.11.16

5 Prometheus. EXPORTERS AND INTEGRATIONS EB/OL.

Blue whale wisdom cloud

This article is edited and released by Tencent blue whale Zhiyun , Tencent blue whale Zhiyun （ Short for blue whale ） The software system is a set of systems based on PaaS Technology solutions for , Committed to building an industry-leading one-stop automatic operation and maintenance platform . At present, the community version has been launched 、 Enterprise Edition , Welcome to experience .