The site and determining its future place in the provider’s server, hosts or other companies, the author is ready to immediately put it out there and give him access to all comers. But it happens that, having made this step, the author notes that his site produces an error, and responding slowly to requests that are not available in the most unexpected situations. There is a need to define what prevents the normal operation of the project – external causes or internal, depending on how well and properly it is written. Allow to test different programs on one of them – WAPT – today we’ll talk.
There are two main ways of testing – both manual and automated. For manual testing is necessary to develop a program of tests to select testers, train them to perform the necessary work, and yes even to repeat it several times to achieve the desired result. Automated testing eliminates many of these problems, leaving the owner’s site (since we’re talking about testing sites) definition of the testing program. Application of the program allows you to automate WAPT repeatability tests and getting results, reduce the influence of the "human factor" of his mistakes, a lot of "virtual testers on the same computer.
The number of virtual testers that run on one computer, depending on its configuration. If you have a configuration is not worse than the Intel Pentium III 800-1000 MHz, 256-512 MB of memory, the number could reach thousands. But it is only in the case of an operating system such as Windows NT/2000. For a family of Windows 9x the number of virtual testers is limited to one hundred, which is regulated by the peculiarities of the operating system. Number of virtual users is restricted and the amount of computer memory. Tentatively, we can assume that for a single user in the process of the program allocated 100 kilobytes for a report and the same – for logging work. In this case, thousands of virtual users will need at least 200 megabytes of memory. The number of virtual testers can reach thousands on a single computer.
WAPT – program for testing web sites and desktop applications with web interface under load, simulating the simultaneous operation with a resource of several hundreds or thousands of users. The program allows you to analyze the characteristics of the application, identify bottlenecks under various loads. The program simulates the actions of actual users of virtual users, where each separate virtual user emulates the real one. Actions that need to perform are described in the test scripts. It should be noted that the program ensures that the testing sites, not only when connected directly to them, but when working through a proxy server using secure connections.
The program interface is simple and straightforward. There are four main sections – the window of recording the sequence of test execution window script editor window and run the test browser testing results. In the writing test will record the sequence of actions to be performed by a virtual user when visiting a site. These actions are performed while viewing the site the browser window (use MS IE). After completing this process, the resulting sequence of actions may be edited. And not only sequence but also add dynamic characteristics of virtual users.
In testing, we consider three possible scenarios – testing under load, testing, performance testing and stress situations, during which time trying to discover the limits of operation and bottlenecks. Another possibility – testing in the background, where it is possible to simultaneously work on the site and as a real user, which gives a more accurate picture of the site.
And in order to test still hold, it is necessary to develop a script. It starts with defining the basic problem that you set for yourself. This may be a check on the maximum number of downloads of one page per unit time, the definition of threshold for rejection, and other tasks. Next, define the process of test execution. In two ways – testing of a certain number of virtual users you have, or testing from one to the maximum amount you set (the so-called batch mode). Use the following installation, you specify the amount of testing – or the number of times each test user, or the duration of testing.
When preparing the job for testing, you can choose the level of realism "of" virtual users. Lies in the choice of setting the connection speed of users and their level of "intelligence." At the highest level you can set random values ??of time intervals between repetitions of tests and setting the speed of access. With an average level – only the speed. At the lowest available, no further adjustment. Thus it is necessary to remember the main rule – the higher the level, the less load on the server.
After you configure the basic functions of a script, you can move itself to the creation of a test script. To do this, open the window Recorder and start recording. Introduce the first address, the links go to other pages that perform certain actions – all of them are recorded and stored in the log work. After you have options, you can stop recording and go into the editing window scenario. On this page you configure the funnel (introduced sequence can be changed), a number of other settings. These include:
After editing the script you can save it and go to the tab Runner. In the opening window is proposed to make additional configuration script – determine the type of test (normal or batch) and within the type to make additional settings. These include choosing the number of testers, test time or number of repeats, intervals between repetitions and a few other options. Among them – can limit the number of queries per second, set the parameters of the report, determine the delay between the start of testing of individual users. The latter option allows a choice between a simultaneous start and work with a delay, which must put its value in milliseconds.
Another group of switches provides a choice of synchronization. Three of them. At the first treatment is over elapsed time between clicking on the link and download the page (Web transaction time), in the second – on page load time (from receipt of response from the server before the download is complete), the third – on-time response to a request (from the date of request prior to confirming the response from the server). In general, the time when the first version is equal to the amount of time on the second and third options. You can also choose to reset the query above a certain you maximum time on request. Otherwise, you can get an infinite timeout when the server fails.
When you have finished configuring your possible move to the process itself. If, however, save the script, it can be later used for re-testing, and to create new scripts based on it.
Well, after a test is executed, you can go to the most interesting part – the analysis of results. To do this, move to a new tab – Results. In this tab, you have three options – to view the results in text format, graphics and graphical interface for batch mode. The generated report can be saved in xls and subsequently loaded into Excel (this is an additional option to the core – Conservation in the format of the program).
The test report consists of several sections. The first displays general information – how and what is being tested, with what settings. Next is a section which listed all the addresses of pages on which testing, with a time delay for the transition to a request from the next page. The next section contains the main results of the test – the minimum, average and maximum response time when accessing each test page. Network statistics show the volume of incoming and outgoing information, as well as average speed. They are followed by information about virtual testers, the number and type of errors, load time. After trying to test a small site discovered that I could not determine the causes of errors. Errors are, but what are they, because of what they have occurred? (This same report can be generated not only in text form, but also in the form of html-pages.)
On page graphical reports generated information on several indicators – the time of the query (after sending the request until it receives the last byte of the requested page), the response time of the server and page load time. A graphics window batch testing make it possible to view information related to the peculiarities of this type of test.
WAPT is very successful tool that is simple to configure and use. Allows you to test the functionality of a Web resource, both in real its performance, and locally, helps determine the thin places, understand the capabilities of the resource by keeping the load of visitors. The only pity is that the program developed by Russian developer – the company Softlogica LLC – has Russian interface and Russian help file.