Response time in performance requriement

转载 2007年09月15日 15:08:00

The most important test scenario for the enterprise:

Maximum number of concurrent users successfully accessing a database with 3 seconds response time. 

3 seconds is a standard “acceptable” response time for modern distributed enterprise network applications.

1 second is “excellent”,

10 seconds is “maximum allowable”.

Concurrent users” have a behavior: 2-3 events (mouse clicks) per minute. 20 seconds per decision made. 


High (good):

Up to 5 seconds



From 6 to 10 seconds


Low (poor):

Over 10 seconds

Acceptable Computer Response Times


In a well-designed website, how long should users have to wait for pages to download?

Past studies:

Using the research that was available to them, Miller (1968), Bailey (1982) and Shneiderman (1984) recommended that the computer should respond to users within two seconds. Martin and Corl (1986) reported that with most data entry tasks there was no advantage of having response times that were faster than one second, and found a linear decrease in productivity with slower response times (from one to five seconds). With problem solving tasks, which are more like Web interaction tasks, they found no reliable effect on performance up to a 5-second delay.

With websites we have had all kinds of interesting maximum "wait time" numbers proposed over the last few years. Most were based on educated guesses, and most have proposed a number (e.g., 8 or 10 seconds) that took into account that loading Web pages would be much slower than the "two seconds" suggested in earlier studies. So how slow is too slow?

divider line
Many issues are involved:

One reason that acceptable response times are so difficult to determine is that people's expectations for acceptable response times differ from situation to situation. Users seem willing to wait varying amounts of time for different types of interactions. The amount of time a user is willing to wait appears to be a function of the perceived complexity of the request. For example, people will wait longer for requests they think are hard or time-consuming for the computer to perform.

Paula Selvidge and Barbara Chaparro at Wichita State University, and Gregory Bender at IBM, conducted a study to examine the effect of download delays on user performance. They used delays of 1 second, 30 seconds, and 60 seconds. They felt that longer delays (30 or 60 seconds) would increase frustration, and decrease task success and efficiency. They found that users were less frustrated with the one-second delay, but their task success and efficiency were not affected by either the 30 or 60 second response times.

divider line
Acceptable wait times:

One of the best recent series of studies was conducted by Anna Bouch (University College - London), Allan Kuchinsky and Nina Bhatti (Hewlett Packard Labs - Palo Alto). They attempted to identify how long users would wait for pages to load.

Users were presented with Web pages that had predetermined delays ranging from 2 to 73 seconds. While performing the task, users rated the latency (delay) for each page they accessed as high, average or poor. Latency was defined as the delay between a request for a Web page, and totally receiving that page.

They reported the following ratings:


High (good):

Up to 5 seconds



From 6 to 10 seconds


Low (poor):

Over 10 seconds


In a second study, when users experienced a page loading delay that was unacceptable, they pressed a button labeled "Increase Quality." The overall average time before pressing the "Increase Quality" button was 8.6 seconds.

In a third study, they had the Web pages load incrementally with the banner first, text next and graphics last. Under these conditions, users were much more tolerant of longer latencies. The test subjects rated the delay as "good" with latencies up to 39 seconds, and "poor" for those over 56 seconds.

divider line
Negative impressions:

Negative impressions of a website, that seem to have little to do with waiting, may be related to how long users must wait for pages to download. Ramsay, Barbesi and Preece (1998) examined the effect of page loading delays on users' perceptions of websites. Their delays ranged from two seconds to two minutes. Users rated pages on "interesting content," and the difficulty with which the page could be scanned. They found that pages with longer delays (41 seconds or longer) were rated as less interesting and more difficult to scan. In another study, participants felt that the slow loading of Web pages suggested that

  1. products being sold were of inferior quality, and
  2. the security of their purchases may be compromised.
divider line

Even after years of research, the complex relationship between computer response time and user performance and satisfaction is not totally clear. It seems that acceptable downloading times can range from five to over 30 seconds. The delays that are "acceptable" seem to depend on what tasks users re performing, and the difficulties they believe the computer is encountering. For example, users will tolerate only short delays if they believe that the task should be quick and easy for the computer.

Slow computer response times:

  1. may reduce the amount of work that users can do,
  2. probably have little practical effect on errors, and
  3. can be frustrating.

As users interact more with a website their frustration with downloading delays seems to accumulate. In general, the longer users interact with a site, the less delay they will tolerate.



Bailey, R.W. (1982), Human Performance Engineering (1st Edition), Prentice-Hall: Englewood Cliffs, NJ.

Bouch, A., Kuchinsky, A. and Bhatti, N. (2000), Quality is in the eye of the beholder: Meeting users' requirements for Internet quality of service, CHI 2000, 297-304.

Miller, R. B. (1968), Response time in user-system conversational transactions, In Proceedings of the AFIPS Fall Joint Computer Conference, 33, 267-277.

Martin, G.L. and Corl, K.G. (1986), System response time effects on user productivity, Behaviour and Information Technology, 5(1), 3-13.

Ramsay, J., Barbesi, A. and Preece, J. (1998), A psychological investigation of long retrieval times on the World Wide Web, Interacting with Computers, 10, 77-86.

Selvidge, P.R., Chaparro, B. and Bender, G.T. (2000), The world wide wait: Effects of delays on user performance, Proceedings of the IEA 2000/HFES 2000 Congress, 1-416-1-419.

Shneiderman, B. (1984), Response time and display rate in human performance with computers, Computing Surveys, 16, 265-285.

The Pragmatic Ergonomist


Bob's article reminds us that our current web technology is only the beginning. It is painfully inadequate in many ways. But, to be practical, we must work with that technlogy. To this end, remember that it does NOT matter so much how long a page takes to load. What matters is how long before the user can begin productive work. Therefore, work with the technology to get the user working fast.

When you insert an image, include a height and width flag in your HTML code. This makes the browser draw a box for the image and display text first. Get the user reading and THEN display the image. Also, use methods like pre-loads. If the user is going to spend some time on the home page, add images that will be used further down in the site. Make these images 1 pixel by 1 pixel so they are not seen. But they will be in cache and will appear to open instantly when the user goes to the next page.


Jmeter 插件说明

1、jp@gc - Actiive Threads Over Time:不同时间活动用户数量展示(图表) 2、jp@gc - AutoStop Listener :自动停止监听器 av...
  • Happy_hai
  • Happy_hai
  • 2015年06月30日 16:06
  • 2962


一、Write方法   Response 对象的Write 方法可以向发往客户端浏览器的HTML 文件中直接输出信息。也就是说,Reponse 对象的Write 方法将信息从 Web 服务器输出...
  • wind8195
  • wind8195
  • 2016年09月18日 12:09
  • 776


 come from: 18.3.12 Aggregate ...
  • zm_21
  • zm_21
  • 2015年07月09日 14:27
  • 4510

CentOS 6.5下安装VNC(tigervnc)服务的详细过程

一、搭建vnc服务器的过程 1.查看tigervnc是否安装。 [root@localhost ~]# rpm -qa | grep tigervnc tigervnc-1.1....
  • xxin1994
  • xxin1994
  • 2016年01月24日 15:55
  • 2811

Google优化工具Timeline的使用(Chrome 57已经改为performance(性能模板))

前期学习知识搭一下页面但是现在越来越注重网页性能的优化。 都推荐TimeLine所以开始使用performance进行性能的调试...
  • qq_26878975
  • qq_26878975
  • 2017年07月07日 10:53
  • 3302

浅谈浏览器对象——Performance API

为了得到脚本运行的精确耗时,需要一个高精度时间戳,传统的做法是使用Date对象的getTime方法,其不足之处在于: 1) getTime方法及Date对象的其他方法只能精确到毫秒级别,无法得到更小的...
  • zhouziyu2011
  • zhouziyu2011
  • 2017年04月25日 09:49
  • 1064

Performance — 前端性能监控利器 最近在写一个监控脚本,终于有机会接触到了这一块,整理后写下了本文。 Performance是一个做前端...
  • z69183787
  • z69183787
  • 2017年11月12日 19:08
  • 120


  • ouyanggengcheng
  • ouyanggengcheng
  • 2017年03月16日 11:07
  • 362


如何在浏览器中查看当前页面请求各个指标数据 在控制台中输入window.performance.timing(html5的属性);即可            各字段的含义: ·    ...
  • q_12341234
  • q_12341234
  • 2015年01月06日 16:46
  • 1726


  • ImagineCode
  • ImagineCode
  • 2016年03月12日 09:50
  • 1463
您举报文章:Response time in performance requriement