December 18, 2014
During the holiday season, shoppers flock to online websites to perform their holiday shopping. What a lot of websites don't realize is that they must perform adequate testing in order to insure that the site can handle the larger traffic generation.
As the time for the holiday season reaches its climax, the shoppers are flocking to finish up their shopping lists. I am proud to say that I am not amongst the mobs at the mall, as I sat down on Saturday morning with my laptop, determined to start AND finish buying gifts for everyone in one day. Although this was an ambitious quest, I succeeded! We are so lucky to live in this technological era, where virtually everything is available to buy online. In fact, IBM reported that Thanksgiving 2014 online sales were up by a whole 14% from last year’s sales!
With the huge rise in online shopping during the holiday season, it becomes quite clear which websites have prepared versus those who have not properly tested their websites properly with the forethought of last minute shoppers. Websites without proper consideration can be obvious due to lag, bottlenecks, functional problems and even just basic availability.
The issues encountered by these faulty websites can be avoided by the proper use of load testing. Load testing is the process of putting stress on a website, system or device to measure its response. It is performed in order to determine how the system will behave under normal and anticipated peakload conditions. For example, a good load test would include the expected concurrent number of users on the application performing a specific number of transactions within the set duration.
The most important key to load testing is to have a realistic model of how your website will perform in the real world, and model the conditions that your site will experience. You must ensure that your testing of the application will make the website react in a similar way as to the peak load conditions.
In order to perform load testing correctly and accurately, the following aspects should be considered:
- Response time under different load levels
- Maximum clients
- Memory Leaks and database lockups
- Client-server throughput
- Memory and CPU usage
- Performance on several data volumes
- Load balancing
- Stress testing
- Screen refresh rate
Poor performance and slow reaction times will result in negative consequences on companies. This could be decreased sales in web commerce, customer abandonment, harmful reputation and wasted resources. Plus, nobody wants holiday shopping to last any longer than it must – it’s stressful enough as it is!