Correct monitoring settings make your Web Services indestructible

My enterprise webpage service must be in the monitored status every day or even throughout the lifecycle. A monitored service must meet one of my 12 operational preparation principles.

If you are properly monitored, you will be able to prove that your service level is perfect, and you will be able to track the error information before the customer notices it to solve the problem more quickly, in addition, you will have a complete set of equipment for measuring system performance.

Everyone has clear expectations for the enterprise services they enjoy. In the design and construction phase of a project, it is expected that it comes first from business needs, the processes that are constantly familiar during the project, and a formal SLA from a fixed SLA ). During its service life, it is expected to be adjusted as the application is updated and provide a better customer service.

End user experience. What does the user see? Is a customer satisfied with business processing?

What happened behind the scenes. What happened behind the scenes? How do these application components work? Do I have enough resources to keep these applications running?

What happened to the client?

Monitor front-end services using a virtual terminal monitoring application. The virtual end continuously executes commands and collects data. This monitoring application records the request commands sent by these users to the Web Service and the responses they receive. Within the scope permitted by SLA), this monitoring application can monitor these response times and draw these response time charts for management reports.

This monitoring application runs a simple monitoring application on the home page to ensure that the home page is not infected with viruses. This application can also perform complex processing to check whether all components of the application work properly.

It is good to monitor an application from a remote client, which can clearly understand what kind of health check a user needs. If a customer thinks this performance is spam, the customer manager will use some measurement functions of the monitoring to respond to the customer's opinion. Remote monitoring also exposes some vulnerabilities. Can I use CDN to accelerate the service of remote users? Can problems caused by ISP Internet service providers or cloud services be reported?

Unfortunately, it is impossible for many customer office vendors to install monitoring applications, so they serve companies with a wide network. In the early stages of the monitoring service market, I used a free Monitor. us application. In the high-end stage, I use Compuware products that provide a wide range of IT software, services, and best practices to ensure the optimal performance of technical products around the world.

What happened on the server?

A monitoring application uses all operating system metrics to detect a backend service. Each cloud service-based application is built on thousands of virtual translation components.

I want monitoring to detect the hardware layer, network layer, operating system, and applications. These must be monitored. Cloud services have multi-tenant and multi-site features. The performance of my applications will be affected by other tenants.

I want to detect all application components. These applications may be distributed in several layers-maybe in the front layer that the customer can see, most of the business logic is in the middle layer, and the data source is in the backend. These must be monitored. I want to know if all components will be affected, for example, when the external force is applied, the site traffic increases, or when a group of inefficient code programs are updated, the database sounds. I installed cacti's network traffic monitoring graphical analysis tool to check the reliability of the operating system when I was working.

All remote services connected to the application must also be monitored. No Enterprise Service is an isolated island. All backend integration needs to be monitored. I cannot stop someone from slowing down the program, but I can collect these metrics and get rid of them.

What will happen over time?

Collecting these trends for evaluation can expose many problems and prevent them.

Traffic usage will change over time. If the service is attracting more customers over time, the owner will be satisfied. If the attractiveness gradually declines over time, the owners will be dissatisfied.

Trend can predict what to send. If more system resources are used over time, someone will know that their bills will increase.

Comprehensive Monitoring

The performance of an application needs to be evaluated, and every kind of thing can affect its performance, including running platform, connecting the application to the customer's network, there are other systems that depend on this application. Monitoring a web service based on cloud services is not just to regularly ping it from your host application.

Creating comprehensive client monitoring is very complicated. Installing an open-source application such as Nagios in your internal system to monitor the system running status and network information is easy, but you need to obtain an appropriate remote monitoring location, you have to use regional distribution service products like Monitis or Gomez.

Creating comprehensive server-side monitoring is very complex, especially for enterprises distributed in different regions. You can connect excellent and free open-source monitoring applications to add them to server-side monitoring, or you can rent out real-time service company products, such as LogicMonitor, New Relic, or BMC.

If anyone expects a service, these expectations must be evaluated to some extent for success or failure. A service must be monitored from both external customers and all internal servers. Monitoring should continue the entire service life cycle.