I've explained more about monitoring, storage, and presentation of a monitoring point in my world of surveillance 1 ~ 4. But in the real world, the whole world is more like a graph, each point can be a monitoring point, and the edge is their call relationship or data flow
Example:
webserver–> MySQL
For one of the simplest web services, it may consist of two parts, webserver and MySQL store stores, commodity information, and webserver services that interact directly with browser users. In such a business scenario, the monitoring points on the webserver may include UV, PV in the unit time, while the monitoring points on MySQL may have connections, requests per second, etc.
In this way, we combine the monitoring point closely with our own business logic. And we can think of each company's business is completely different, so if we put the monitoring also level, we found that monitoring is also hierarchical
3 Business-level monitoring |
2 Monitoring point data collection, storage, aggregation |
1 Monitoring Point Data acquisition |
1) Monitoring point data collection can not take the passage scheme
In addition to basic types of monitoring such as CPU utilization, load, disk usage
But similar to UV, PV business indicators are difficult to unify together
2) Monitoring point data collection, storage, aggregation can be used to solve the traffic plan
Recently I have seen an article that uses ES to do post-aggregation, without first aggregating
3) Business level monitoring to reflect the business flow of the various applications, services between the call relationship, especially to reflect the flow of data, the size of traffic and so on. Data flow is the core element throughout each module
Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.
My monitoring Worldview (5)--how to reflect business scenarios in monitoring