如何在没有网络延迟的情况下获得webtest的响应时间 [英] How to get response time of a webtest with out network latency
问题描述
我正在加载测试，我的客户有如下要求，
i am load testing and my client has a requirement like below,
- 获取每个请求的响应时间，包括网络时间
- 获取每个请求的响应时间（不包括网络时间）
根据我的想法，响应时间在加载测试结果，包括网络时间..有人可以告诉我如何计算请求的响应，包括网络时间。
As of my idea, the response time in the load test result, includes Network time.. Can some one tell me how we can calculate the response of request with out including Network time.
谢谢，
Joshi
Joshi
推荐答案
影响network 延迟取决于与服务器所在位置相比产生负载的位置。要将网络延迟减少到尽可能小的值，请将负载生成器（代理计算机）放置在物理上靠近
服务器的位置，并使用非常高带宽的网络链接来连接它们。物理上接近我的意思是尽可能少的干预网络路由器或交换机等。如果服务器是云托管的，那么将负载生成器托管在同一个云中。
（我的意思是相同的云。如果可能的话，只需在服务器旁创建一些额外的虚拟机并使用它们来运行负载测试。）
The impact of network delays depends upon where the load is generated compared to where the servers are located. To reduce network delays to as small a value as possible, locate the load generators (the agent computers) physically close to the server and have a very high bandwidth network link to connect them. By physically close I mean with as few intervening network routers or switches etc as possible. If the servers are cloud hosted then have the load generators hosted within the same cloud. (I really mean the same cloud. If possible just create some extra virtual machines beside the servers and use them to run the load test.)
另一种可能性非常好的东西类似于客户想要的是测量网络时间，然后从总时间中减去它。通常，被测系统由许多计算机组成。其中少数计算机将连接到
外部网络，他们将接收来自负载生成器的请求并发送回复，但其他计算机上内部网络将完成大部分工作。使用可直接连接到
外部网络的计算机处理的请求。 考虑一下 "echo" 命令，有点像"ping"一样。想要在外部服务器中具有最少处理并且发送响应的请求。现在，您可以测量处理实际请求的总时间
以及处理简单回声的时间。不同之处在于被测试服务器的主要部分所花费的时间。
Another possibility that gives something very similar to what your client wants is to measure network time and then subtract it from total time. Commonly the system under test is made of many computers. A small number of them will connect to the external network, they will receive the requests from the load generators and send the replies, but other computers on an internal network will do much of the work. Use a request that can be handled by those computers that connect directly to the external network. Think of an "echo" command, a bit like "ping". Want a request that has minimal processing within the external facing server and that sends a response. Now you have measures of the total time to process a real request and the time to process a simple echo. The difference is the time that is taken by the main part of the servers being tested.
还有一件事。问你的客户他们为什么想要这两个值？他们期望用这两个值学到什么？他们是否在一个地方为世界各地的客户提供服务？等等......
One more thing. Ask your client why they want the two values? What do they expect to learn with these two values? Are they serving customers at many places around the world from a single location? And so on...
问候
Adrian
这篇关于如何在没有网络延迟的情况下获得webtest的响应时间的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
