Shulou(Shulou.com)06/01 Report--

I have been engaged in the field of software testing and quality assurance for more than ten years, from white box automation testing to black box automation testing, from function testing to performance testing, safety testing, from test flow, test framework to later test management, quality system construction, deeply love this industry, often excited for conquering a technical hill with small partners, but also for working overtime to complete a difficult task.

I have been responsible for testing technical management and team management for more than 8 years, during which there are at least one or two thousand people interviewed. Here are some common questions from interviews to talk about the basics of performance testing:

I. Basic articles

1, a relatively complete performance test process

2. Basic theory and common terms of performance test

3. Performance test model and type

4. HTTP and TCP protocol related knowledge

5. Connection pool and thread related knowledge

II. Tools

1、Jmeter

1. What is the working principle of Jmeter?

②. What are the commonly used components and plug-ins? What are their respective roles?

③, several typical scenarios, how to design test scripts based on jmeter?

For example: parameterization, association, TPS control, interface encryption signature, stepped pressurization, assembly point, checkpoint, etc.;

Will it be redeveloped? If yes, how about secondary development (introduce the approximate process and reasons)?

2、Loadrunner

3. Other open source/commercial performance testing tools

For example: Ngrinder, Locust, WRK, Artistry, etc.;

4. Front-end, server and database performance monitoring tools

III. System Architecture

1. Service cluster

2. Load Balancer

Principle and Implementation of Load Balancer

3. Capacity planning

4. Cache application

Cache Principles, Cache Benefits, Cache Hits, Cache Penetration, Multi-Tier Cache

5. Distributed framework

Characteristics and challenges of distribution: CAP theory (data consistency, service availability, partition fault tolerance)

6. Full link voltage test

About Full Link Voltage Measurement

Server & Middleware

1、JVM

JVM Principles, Startup Parameter Configuration, Stack Principles, Garbage Collection Principles, OOM Causes and Manifestations

2、Tomcat

Configuration, usage, startup parameter configuration

3、Nginx

Configuration, usage method

4、Dubbo

Service Registration, Message Queue

5、RabbitMQ/Kafka

How to manage its own characteristics, producers and consumers

V. Database

1, lock

2. Index

3. Separation of reading and writing

4. Sub-warehouse and sub-table

VI. Program

1. What questions need to be considered in designing performance test schemes?

Time cost, labor cost, environment & script reusability, implementation difficulty

2. How would you design and optimize the plan for some situations?

VII. Cases

How to test MQ?

2. Analysis of the reasons why TPS does not go up in pressure measurement?

3. How to choose the ratio of test environment and production environment server?

Server configuration versions remain consistent, capacity tests are followed by equivalent substitution, marginal diminishing effects are considered, and disaster recovery schemes are implemented.

4. How to analyze bottlenecks?

Bottleneck analysis granularity from top to bottom, from local to global

5. How to prepare test data? How to prevent data contamination?

Production data backup, data isolation, test data falling into shadow library, baffle, mock

6. Common methods for performance optimization

Consideration on Space-Time Reliability of Performance Optimization

VIII. Summary of common 4 types of problems

Question 1: In what environment should the interface test be conducted?

If the test equipment allows, try to deploy the test application to the test server. And build the test environment according to the deployment mode of production environment.

If applications that interact through interfaces in production are hosted on different servers, test environments should also host applications on different servers. The production environment uses vpn, and the server in the test environment is preferably also in a vpn environment.

Question 2: Can the stress testing tool be installed on the tested WEB server for local stress testing?

The general requirement is to separate the pressure and the pressed machine, so as to facilitate the analysis of the pressure test results. It is best to place the pressure test and the pressed machine in the same room, so as to reduce the data fluctuation caused by the network and bandwidth and the bottleneck caused thereby.

Whether it was pressing or being pressed, it would take up resources. When system bottlenecks occur, if both are on the same machine, it can have a big impact on your analysis.

If your upper layer request is the same idc call method under the same intranet, then you can choose the same idc machine for pressure reference.

Of course, if your expected QPS is very low, it will not reach the bottleneck of the machine at all, just verify some simple concurrency problems such as memory leaks, and mix it up.

Question 3: Performance testing of cloud platforms

The performance test of cloud platform is similar to that of general performance test. The performance concurrency capacity of cloud platform is relatively large, and the number of users is a little more.

1. First, analyze performance test points such as: occurrence frequency is very high, critical degree is very high, resource occupation is very serious, etc.

2. Performance indicators such as: home page open speed, maximum online users supported, number of concurrent users supported by the system, etc.;

3. Choice of tools: JMeter, Loadrunner, etc.

Question 4: How to set up to keep users always online;

1. Record different scenes in multiple scripts and then concurrent controller should be on the line.

2. After logging in, do nothing, must be time-out of the line. Scripts that stay online should periodically send requests in actions that make the server think you're still online, such as heartbeat requests that would otherwise drop the connection.

Another way is to keep the thinking time unchanged and not send anything, but let the developer modify the server-side timeout to become very long, which can also achieve the goal of not dropping the line.

Summary:

I believe these common interview questions will be encountered by many people in practice. As a software testing engineer, rigorous interview questions, pay attention to improve their good communication and expression skills, in the interview process reflects a solid software testing basic knowledge, sufficient patience, carefulness, confidence, responsibility, good at self-summary, self-supervision and continuous learning ability, I believe everyone can have excellent performance in the interview, find the desired job.

Welcome to join 51 software testing family, here you will get [latest industry information],[free testing tool installation kit],[software testing technology dry goods],[interview job skills]... 51 Learn and grow together with you! Looking forward to your joining: QQ group: 755431660

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.