Analyzing Response Time Bottlenecks in Web Applications

Abstract: A common problem amongst the millions of web users around the world is the performance of web sites. Performance in common man’s terminology can be defined as the time experienced by the user for an action performed. Action could be as simple as a log in to an application or as complex as creating a complex report (one that has a lot of data dependencies). All the user sees on doing a particular action is an hour glass rotating or a message stating that “Processing is in progress; Please wait”.
What is Response Time? Response time in performance terminology is the time taken by the server to respond to the request. In end user terminology, Response time is the time for the user to view the data requested on the page. The industry standard for viewing data on a web page is 3 seconds. More often than not users would stop or discontinue the activity on the browser if it takes more than 3s.
Why do we see such high response times in our applications? This is the question that users tend to ask performance engineers. In the subsequent sections of this paper we would discuss and analyze some of the probable causes of high response times which translate to low performing applications.


Profile: The author has been with the Performance engineering team at Symphony Services for the past four and half years.

He has more than 10 years of IT experience of which more than 5 years have been spent in performance engineering of different applications. All the case studies mentioned here are real experiences observed by the author in the performance cycle of the different applications.