Micro Services practice: operation and maintenance services

Monitoring and alerting

　　Monitoring basic objective is to control the health service creation environment in a timely alarm after a system failure, and the ability to quickly locate problems by monitoring information . Another goal is to monitor fault warning, perception before a failure occurs according to the rules set in advance and notify maintenance personnel, or automatically make operation and maintenance decisions.

Monitoring indicators involved

Server Resource Monitoring : CPU, memory, disk IO and other information.
Service Performance Monitoring : The most basic is the response time monitoring service.
Business Monitoring : Monitoring systems can also understand the status of the business. For example, by monitoring the message queue, the order can be seen that the number of currently processed; monitoring the test environment generation of A / B, it is known that characteristics more popular with users. Monitoring of the business, allowing decision makers to better grasp the status of operations and on the basis of health data, make more accurate business decisions.

Monitoring strategies micro services

　　For micro-services architecture system, which monitor typically consists of four modules: data collection, data aggregation, data storage, data display .

　　Monitoring information collection module data collection and monitoring indicators sent to the data aggregation module. Here's index will collect a wide variety, such as the need to collect components for servers, databases, service interface, service dependencies, such as monitoring information ElasticSearch respectively. Next, the data aggregation module when data is received, the polymerization further, collected in the process between the serving and the processor may be processed by the asynchronous message queue. Then, the data aggregation module relating to the data packets after polymerization into different indicator database, storing the processing result to the database. Considering the particular nature of data and the amount of indicator data, the database may select the time-series database. Finally, the data display module reads data from the database, and display. It also provides a wealth of indicators name search tool.