网站性能测试
终极指南

Performance Testing Best Practices

We’ve discussed why performance testing is important. Your customers, users, and visitors all want a fast, uninterrupted experience. Slow load pages and applications that are difficult to navigate are going to make them turn away. And that’s going to result in unhappy users and losses in revenue that will go directly into your competitor’s wallet. If your website or application is going to be used and accessed by a lot of hundreds of thousands of users simultaneously, you’re going to need to understand how it performs under those conditions. Marketing promotions can intentionally, or unintentionally, drive a lot of traffic to your site, so performance testing under a variety of conditions, spikes in traffic, and from different geographic locations is needed to understand the user experience, from the user’s point of view.

Successful organizations understand the necessity of not only just load testing, but also the various functional testing that needs to be executed during the development cycle, such as integration testing, unit testing, smoke testing, as well as others. Any deficiencies found during these tests can help improve performance and alleviate any issues when it gets to the point of performance testing. We’ll discuss these factors and other best practices to consider when carrying out performance tests.

Performance Testing Basics (For Beginners)

We’ve discussed the many reasons for implementing performance testing, the various performance testing tools, and what to look for in a performance testing tool, but now we’ll explain what performance is at a basic level. If you’ve never heard of performance testing or just beginning to investigate performance testing for your own research, we’ll explain what performance testing is, the performance testing process, and different approaches to consider when carrying out performance tests. When developers and engineers talk about performance testing, they’re typically talking about load testing, stress testing, volume testing, or endurance testing, as these are the most performed during the latter stages of application development. However, performance testing also includes compliance testing, recovery testing, usability testing, and more.

Performance testing is a form of non-functional testing. Non-functional testing consists of testing how a system operates, rather than testing how, and if, specific components function. The purpose is to understand how your websites, applications, or APIs, for example, perform when pushed to their limits by simulated users, or load, and how system resources respond. In this way, you can identify how the system behaves, where performance bottlenecks exist, and where improvements need to be made, whether it’s to the application code itself or infrastructure/hardware.

Intermediate Performance Testing

When it comes time to execute performance tests, there are many different scenarios, technologies, and components you can test. For example, you can execute tests against your website, web applications, APIs, and streaming media. Each of these require different levels of set up, execution, and considerations. For example, when it comes to testing your website, you’ll want to consider if you’re going to just test at the protocol level, meaning just pushing simultaneous users against a URL and confirming that your website is available and there are no errors. The benefit of this kind of test is that it can be set up relatively quickly and you can run a high number of concurrent users without using up a ton of resources.

Protocol-based testing is still important and used regularly today, but today’s browsers are more complex and rely on dynamic elements way more than they did even just several years ago. some of the more intermediate performance testing scenarios consist of testing your website or application using real browsers. The benefit of using real browsers for your performance tests is that you can see all the individual elements, third-party components, and code like HTML, CSS, and JavaScript. This gives you total insight into back end server responses, in addition to front-end components that could be affecting performance. Additionally, the closer you can get to testing performance by simulating the way your users and visitors access your sites and applications, the better data and analysis you get back. The downside to real browser-based testing is that it does take more resources and is typically more expensive compared to HTTP performance tests.

Advanced Performance Testing

When it comes to providing a great experience for your customers and visitors, you need to be able to put yourself in the driver’s seat and see how they perceive the performance of your websites and applications. When it comes to advanced performance testing methods, being able to simulate the experience by creating scripts that emulate critical user flows and scenarios, and then testing those scripts against high levels of concurrent and simultaneous users from multiple points around the world. We talked about the many performance testing tools and some of them provided more user-friendly scripting options. Some of the on-premises solutions and tools require an in-depth knowledge of specific technologies, while some, like LoadView and LoadNinja, offer a point and click scripting tool, that require no previous scripting experience.

However, having scripting experience is important when it comes to advanced performance testing techniques, such as configuring and customizing your scripts. It’ll make the whole process easier and less time consuming. For example, some tools allow you to set delays in between actions, clicks, and text speed, to better match user behaviors. Some tools may give you preset values, but you may want to set these accordingly to match your visitor’s actions and set them manually. Other advanced methods include how to set up your load curves for your test. Some tools allow you to just set the maximum number of users over a given period of time, but some solutions, like LoadView, give multiple load curve options, including an option to increase or decrease load levels during a test to see how your system responds in real-time.

Performance Testing Benchmarking

As part of the performance testing process, all software and applications should be tested to make sure it satisfies performance and business requirements. Part of that process includes benchmark testing and baseline testing. Even though they sound similar, they are quite different. The goal of baseline testing is to ensure a consistent product. To do this, team will test the software and measure different performance aspects, such as code, network, hardware. The results of the test are recorded and documented. If the software or application is updated, it’ll be tested under the same conditions to see how the results compared with the previous test.

Benchmarking is a bit different. Benchmark tests essentially compares performance of an application against other applications or an industry standard to ensure it meets or exceeds quality stands. This is especially important for organizations looking to set quality standards or meet specific service-level agreements (SLAs) of their applications and software for their users and partners. Benchmarking is very much driven by the business and organization, helping to provide trust with potential customers and showcases your organization as a leader in your space.

The Performance Testing Life Cycle Explained

As we previously described, performance testing consists of testing the non-functional aspects of a software application, like application behavior, usability, and reliability as well as understanding how the system responds and how resources are used. The performance testing life cycle (PTLC) consists of different phases and strategies. Let’s dive into the various components of the performance testing life cycle.

The first and most critical steps of any kind of software testing is ensuring that the system that’s going to be tested is complete. Performance testing essentially begins where functional testing left off. It’s essential that the software and system functions, so you can accurately test and measure results.

The next step is to come up with the test strategy. This includes preparing the scope of the test, the workload models, and identifying the scenarios that will be tested. In addition to that, teams will define the metrics that will be used to measure success, however, sometimes this will change once the tests are executed and data is reviewed. It’s possible that something unexpected happens, which would affect the testing process.

Next, it’s time to design the test itself along with execution. Depending on the performance testing solution you’re using, this may occur at the same time or might be separate steps. This step consists of creating the test scripts and user actions that were defined in the test strategy. Next those scripts are prepared for performance testing against the agreed upon workload models.

Lastly, once the performance test has finished, performance engineers can view the test data and analyze the results and begin the process of making recommendations for improvements. It’s also here where baselines are set, so any additional performance tests can be used to compare results.

性能测试优点和缺点

对您的网站、Web 应用和 ABI 进行性能测试对用户体验至关重要。 确保您的网站和应用程序高效、快速地运行，即使在流量很大的情况下，也是实现支持业务目标的创收的关键。 正如我们在上一节中谈到的，性能测试可以帮助发现功能测试无法检测到的问题。 例如，加载速度更快的网站或应用程序比加载速度较慢的页面或应用程序具有更好的保留率。 遇到加载缓慢页面的访客可能再也不会重访该网站。 第一印象就是一切，所以很明显，性能测试有很多好处。

虽然性能测试有许多优点，但它确实存在一些缺点。 一般来说，任何类型的测试，有一个成本和时间投资，需要考虑。 并非所有团队和组织都是一样的，因此某些团队可能需要付出更多努力。 然而，市场上许多基于 Web 的性能测试工具可以帮助缓解这一问题。 与许多本地工具（往往更昂贵且需要技术知识）不同，基于 SaaS 的性能测试工具可消除许多耗时的问题，例如设置自己的硬件/基础设施和管理许可费。 最后，性能测试可能会发现需要额外系统容量和升级的意外问题，最终成本可能超过您计划的成本。

性能测试清单

通常，一旦功能测试完成，应用程序接近软件开发生命周期的结束，下一步是在特定工作负载下测试应用程序和功能，看看您的应用程序和基础系统如何响应。 在更敏捷的开发过程中，性能测试定期完成，以检查整个开发生命周期的性能，因此，当应用到达开发结束时，大多数瓶颈和问题已经得到解决。

为了帮助执行性能测试过程，创建性能测试清单可以帮助定义需要执行的步骤和场景，并帮助您更好地遵循您的计划。 性能测试清单应包括应用程序的考虑因素、客户/用户/SLA 要求、您的系统和环境以及您识别的任何其他内部因素。 记录您的清单的好处有很多目的，例如保持测试计划正常进行，但它也允许您修改并包含测试过程中出现的其他程序和考虑因素。 性能测试清单也可以作为客户端的验证，因此您可以直接显示它们，并引导他们完成您制定的详细计划，这有助于建立更多的信任、提供附加值和验证整个过程。

Production Environment vs. Performance Testing Environment

Traditionally, performance testing wouldn’t be carried out until after QA, which often caused issues, as any performance issues would cause the product to be delayed as bottlenecks needed to be addressed. As development cycles became shorter, it pushed performance testing to start sooner. In today’s DevOps environments, performance is continually validated at the component/code level and then full end-to-end performance testing is once the application is built. performance of until your application or site was pushed live.

Setting up a test environment, also called a sandbox, allows development teams to carry out testing without affecting the production, or live environment. While setting up a duplicate environment does add additional cost and resources, it greatly reduces the risk of unexpected issues arising in the production environment. For example, executing load tests on your website in the live environment could inadvertently cause your test to become a stress test and cause the site to crash. Now, there are some situations where testing is production is ok, but those should best be executed when traffic to your website is light to minimize the impact to users.

Understanding Performance Testing Reports

We’ve discussed in detail most of the aspects of performance testing, but one thing we haven’t discussed is probably the most important pieces of the puzzle: the performance testing reports. These reports, dashboards, and metrics are what are used to understand where bottlenecks exist and what system enhancements are needed to boost your websites, applications, and APIs. Although reports vary for the various performance testing solutions, and some elements may be called out differently, there are still some core performance metrics that need to be inspected and scrutinized in the performance testing reports. Additionally, most tools allow you to easily share these reports with others, so you can receive and gather feedback from various departments.

If you’ve done a good job with planning your objectives and executing your performance test, it will be easy to identify in the performance reports where issues exist. If not, it may just add more confusion and require you to go back to the beginning and reassess your performance tests. Compared to functional tests, where the outcome is easy to identify, since it’s only pass or fail, performance testing is bit more complex and requires additional analysis.

One of the most important visuals that performance testing tools create is the waterfall chart. Within the waterfall chart, there’s a myriad of information and data to be analyzed. For example, one of the most important performance testing metrics to watch for is load time. Remember, visitors are more likely to abandon your pages if the page load time is too slow. In your performance testing requirements plan, if you required your website to load under 3 seconds under the specified user load, the reports will show you that, along with all load and response times of the individual components.

Performance Testing Jobs

What Do Web Performance Engineers Do?

Performance engineering is one of the most complex and demanding positions within DevOps teams. The people within these positions typically have a combination of skillsets, as these jobs require knowledge about the entire software development life cycle, including performance testing, scripting, UI, systems engineering, etc. Someone who is a performance engineer has usually had experience in other positions or backgrounds, like QA testing, coding, or network/database administrator, for example. It’s a position that requires a diverse set of knowledge and experience to carry out successful software deployments. A good performance engineer will not only be able to understand how to build a successful product, but also understand the business requirements, so that the product is developed with best practices and industry standards in mind.

The way applications are developed today and completely different than the process was just a decade ago. Development life cycles have become shorter and shorter, partly due to meet user demands, but also to implementing new software development strategies. The performance engineering methodologies and practices are aligned with the Agile process and shift-left testing. Performance engineers are also responsible for suggesting to business stakeholders where and when investments in infrastructure and capacity need to be made, as well as continually tuning and monitoring the system.

Performance Testing Interview Questions

Are you looking for a career as a performance testing engineer? If so, here are some of the top performance testing interview questions you may be asked. Use these as a guideline for the interview process to prep yourself for what you may be asked.

• Can you describe a time when you executed performance testing and walk me through the steps you took?
• What kind of performance testing do you have experience with?
• Have you ever encountered a situation where performance testing was ignored, and it had a detrimental effect to an application? If so, what did you do to remedy the situation?
• What kinds of performance issues did users most encounter?
• How important is the user experience to you? And what does that mean when developing software?
• Can you tell me some of the performance testing tools you have used?
• What performance testing tools did you find were the most successful?
• Performance testing requires teamwork and collaboration. Can you tell me of time that required you to go above and beyond and work with different teams?
• What kinds of performance testing data and reports have you worked with?
• Was there ever a time where an application crashed unexpectedly, and you were called to restore the application? If so, what were the steps you took and were you successful? What was the impact to users or visitors?
• What performance testing mistakes have you made? And how did you overcome those mistakes?