Observable Hybrid Load Testing Scenario

Stress test background

In recent years, in the process of enterprise digital transformation, enterprises have gradually updated and iterated their business systems. In order to ensure the stability of the business system, capacity testing, that is, stress testing, is generally performed before the business system goes online. However, most companies only conduct tests on the intranet through the open source tool JMeter. During the stress test process, they mainly observe the usage of server resources under different pressure values, while ignoring the business itself. As a result, the stability of the business cannot be guaranteed after the system goes online, and various problems occur.

The results of stress testing are naturally unpredictable. No matter how good the stress testing plan is, if it cannot obtain effective data for analysis, the effect will be greatly reduced. Especially in complex and deep system environments, it is particularly important to achieve observability of the entire stress testing process by obtaining full-link tracking data and indicator data.

Stress testing plan

Through Tingyun's observable hybrid full-link stress testing solution, real users' network environments and devices are used to initiate concurrent pressure. At the same time, Tingyun's active and passive telemetry methods are combined to collect user-end, network layer, server and other related indicator data, so as to promptly discover front-end and back-end performance issues and optimize and adjust them in a timely manner.

Front-end pressure: Use the 300,000+ real computer devices, real networks, real operating systems, and real browser users + cloud JMeter pre-deployed by Tingyun in various regions around the world to launch a hybrid stress test of "business processes" and "API interfaces" on the business system.
Telemetry methods: Through a variety of telemetry methods, the business system is made observable during the stress test, so that performance bottlenecks can be discovered in a timely manner and problems can be quickly located.
outputcontent:

Real user experience: Observe key business process availability and performance indicators from the perspective of real users.

Code performance bottleneck of application server: By deploying APM probes, observe the performance fluctuations of the application server and each component.

Resource consumption of the host server: By deploying the Infra (basic component) probe, observe the resource consumption of the host server such as CPU, memory, disk, etc.

Performance of components: Use APM probes to observe database SQL execution, number of database connections, etc.

Observable Hybrid Load Testing Scenario

Stress test background#

Stress testing plan#

Stress test background

Stress testing plan