Standard website monitoring tool and can prove that they are turning the page response, and they will not issue a warning to you. But you can use stress testing technology to monitor your website, run by an interactive script to detect the problem and generates the necessary messages. It is like a quiet sentinel continuous operation, never go to bed or off a fake enhance the reliability of your website.
Our site contains a user logs in, the user authentication process occasionally be stopped. Our standard website monitoring tool can turn on the home page and validate your pages in response, but may interact with the outside is not appropriate to place our tools page. When we warn customers to a real problem that we can only understand it. This is unacceptable, we have to find a better way.
Before we use a stress test tool development and implementation of a series of stress tests, allowing us to run large numbers of users do a lot of different actions on the test site. But we need a way to do simple script repetitive basis up to run a simple user, 24/7, in which the impact on our production system to warn us before we have a real problem customers. Our stress testing tool as a single user to do this test, but when a problem is detected, it lacks a method for generating a warning.
Working with our suppliers, we find that they offer a simple solution: As a single user to repeat the process and have some way to warn of an error when using a different application to perform a stress test scripts. We now have this right processes for three years, and it has an excellent solution. Here is how we operate it.
Design test
The first step is to do some analysis to decide what business to be tested and what failure looks like. When similar stress tests, this test is not only focused on the result of the pressure of time and pages of script to run. You also need the ability to repeat basis using a known username / password combination to log production systems.
The purpose of this test is to simply verify that the site is active and ready to use. Our test does not contain a transaction (sales orders), but you can include this; it just requires more work.
We check these:
- · Each page needs to load in less than 5000 milliseconds (5 seconds)
- · Each page needs to load correctly
- · Each page needs (content verification page load expected) by checking the text
Scripting
Now that you have designed, you can create your script.
First, we choose a legitimate user account can be used for this process. (It needs to be present in the product but is seen as a test account.)
We can also add some of our local area network monitoring stations, used exclusively by authorized users. This authorization through the use of a condom Sockets Layer control, so for those sites, we had to add some special code to support Secure Socket and port mapping.
We use our common stress test script design tool to create a script, a special rule is monitoring tools need to be included in any measure in this design. If you are supported by your support vendor might have some useful information in this area.
Once you have the script was designed and run right, now you can push it to the monitor's application.
Creating Monitors
You need to be covered with your provider to verify what script elements, because in some cases you need all the real-time file, not just the script. The good news is that these files are small.
When you create your monitor, first decide what page elements to be tracked. Depending on your solution, you choose each element can be used to monitor a part of your protocol capacity, so you may choose to limit the number of monitored elements. In our case, we only need to monitor the results page loads and how long it takes to load, but you may also choose other elements, like below some arbitrary.
When you save the monitor, you can see the results of the initial monitoring created on the instrument panel.
Creating monitoring principles
Once you create a monitor, you can why the kind of system used to determine the pass or fail condition to create a rule. We only select the monitor page status and load duration.
When you have a set standard / fail conditions, the reaction state in the instrument panel.
You also need to decide how long you need to test execution. When the first test is still running if the second attempt to start the test, you can get false alarms, so we allowed between each test three minutes.
Creating alert rules
Now you create a test and define what is a failure, you need to create to tell who's as likely to fail to be detected when the rule.
Our tool allows more than 10 different alarms activities, but we have chosen for our e-mail alert process. We use a clear message subject to reaction site problems. We can send an e-mail or text message, if the unit carrier supports e-mail address - for example, [email protected] .
One of the things is to distinguish whether the technology is easy to make mistakes, occasional event does not occur on your website. In order to reduce the number of false alarms, unless the test fails three times in a row, otherwise we do not generate an alert. We also established a rule, every hour to generate a duplicate messages (per 20) only when the price Enforcement Administration is processing an event, they do not need a bunch of e-mail telling them already addressed the event.
We also needed to be involved and alert unavailable to prevent downtime planned provincial capital error message alerts.
Generate reports
These monitoring tools provide ways to check the performance of a good time out and generate useful reports. These can be cut or copied to export to HTML. (I found it was easier to cut and copy the relevant information rather than trying to explain all of the data generated reports.)
These are examples to our quality of sample reports, the login process on Ben collapse and the need for timely changes:
Maintenance Reliability
Our tool also includes a dashboard view, providing a simple place to quickly check all the monitors and see whether there is one or more questions, it may be a bigger problem.
Initially, this tool some strong teams to sell the US Bureau of price controls, because the idea of using virtual user testing is to some people, is a new concept. But now it was like to perform a quiet sentinels, never go to bed or a vacation, we actually have a new application arrival, ask us to detect their sites to ensure that users are able to maintain the reliability.
This improves the reliability of our operations - and this not a quality assurance role it?