Customers will not stand for slow reaction times on a web site. Companies only have a few moments to display a page or complete a transaction before a customer gets frustrated and surfs away to a competitor.
To make sure a company’s web application keeps customers and does not turn them away, a company has to test for the number of customers that may hit a web site.
If you are concerned about satisfying your customers demand for a fast and smooth web experience, then automated software testing provides a way to develop good web application performance.
Lets assume that your web application has been functionally tested, meaning all user-initiated actions produce the expected results. For example, when you want to buy and sell stocks on your online brokerage account you go through several steps. Logging into the account, looking up your portfolio, viewing market data, selecting a stock to sell or buy and lastly completing a stock transaction.
A web application may be functionally perfect (a user can complete his brokerage transaction online,) but under certain load conditions, the user’s experience can become so unacceptable he does not complete that transaction, or worst, the web application cannot serve the user because it is too busy dealing with processing other user requests.
For a web application to perform well, the application must serve all customers at the same time with a minimum standard of performance.
A customer’s web experience depends upon the performance of the following components of a web application: efficient software, well-designed database transactions, properly tuned web and application servers, and network infrastructure all impact performance.
To test the performance of your web application, you will want to understand the behavior of all those components during a load test, or you may not find the bottleneck that was the cause of the bad web performance.
In addition, web applications change over time, for example at the online brokerage, a new portfolio of mutual funds is added. Your company will have to plan for a change such as the number of customers and infrastructure enhancements to the web application.
Options for testing a web application include manual testing for limited user load levels, however, this quickly becomes impractical when reaching higher loads that would require coordinating several hundred people to log onto a brokerage account at precisely the same time.
Another option is using testing software, since the start of the web, programmers have built in-house test scripts to test their applications rather than manually test.
Lastly, you can use commercially available web load testing tools.
Automated tools let you build, record and customize testing scripts. Similar to a Macro in Microsoft Word, or a recording cylinder on an old player piano, these software tools let you record the whole process of completing a stock transaction. Once recorded, you can alter the script to simulate 10’s, 100’s or 1000’s of customers completing a transaction.
You can use the tool to run different testing scenarios, from buying a stock to selling a stock, and run various combinations of those scenarios at the same time in order to simulate the reality of what the online brokerage would face on a typical day.
One example of a tool is QuotiumPRO from Quotium Technologies; their tool allows you to easily create and monitor load scenarios for eleven different databases, web and application servers, and network components running your applications back office.
Other benefits include: accelerating a site launch date by allowing testing throughout development. Increasing the number of defects found during testing. Providing organized, detailed test logs and an audit trail of executed tests, and lastly graphical reports.
Poor web performance can directly hit the bottom-line and will always affect the perception of your company’s brand. There are several automated tool vendors to consider from Massachusetts including: Quotium Technologies, Segue, RadView and Empirix. Others: Mercury Interactive, Rational and Compuware.
Recent Comments