Web application Load-Testing

Web application Load-Testing is a service designed to understand how a web-based application, and/or its supporting infrastructure, behaves under conditions of user load. This load is generated by load-testing tools designed to simulate multiple users accessing the system and performing common user tasks.

Whereas the Application Profiling service allows us to understand the baseline performance characteristics of an application for just a single user, the Load Testing service allows us to take the understanding of performance to the ultimate conclusion, and answer these common questions:

• How will our application perform once deployed in production?
• Will it scale sufficiently to meet user/customer demand?
• What is the maximum number of users the application can serve?
• At what level of load will performance degrade?
• How will the application behave under constant high levels of load or system stress?
• Where will it break? Why?

As a stand-alone service, load-testing can be effective in answering the above questions. However, as many customers with previous experience of using load-test tools and test consultancies will confess to, it can all get very complex, long-winded and very expensive!

Diagnosys can uniquely help here in keeping the test-plan as simple as possible, and as effective and low-cost as possible.

The wrong way to do Load-Testing

The main reason why load-testing projects can become complex, time-consuming and resource-hungry, and sometimes go badly wrong, is because most load-test methodologies employ a complex "scatter-gun" principle;

• Test-scripts designed to test the entire application & all permutations of user activity
• Load-test the entire-application and server infrastructure to try to find performance break-points
• Monitor the entire system for performance characteristics & bottlenecks
• Re-script and re-test ...

Obviously, for a large, complex and widely distributed application system, the implications are quite onerous.

At Diagnosys, we do not believe in this methodology.

The right way to do Load-Testing

We need to keep things simple to ensure success and control costs. The secret, in reality, is quite simple:

• Know where the performance bottlenecks are, and what to test for.

Load-test tools alone are good at informing you as to When application performance will degrade. However, they are not good at identifying the Where and Why, which is usually all part of the same fundamental business question at the heart of the study!

We solve this by integrating the Application Profiling service within the scope of the load-testing project.

Stage 1. Use Application Profiling to baseline the performance of a single user in zero-load test conditions

• Reveals the purest inherent performance characteristics of the application system, free of the effects of network or server load.
• Establishes a performance baseline or reference that future stage test results can be compared to.
• Identifies inherent performance bottlenecks within the application and system that can limit the ultimate response-time performance and scalability when the application is under user load.

Stage 2a.  Use Load-Testing to test the application and server infrastructure at discrete intervals up to the desired maximum number of users

• Reveals how the application behaves with increasing user load

Stage 2b. Use Application Profiling in parallel with Stage 2a to monitor any previously identified performance bottlenecks.

• Rapidly identifies when, where and why the application and/or client-server system is breaking-down and impacting performance and/or scalability.

Stage 3b. Monitor the end-to-end system knowing where the hot-spots are, and what metrics to measure.

We have proven this approach time after time, and can vouch for its effectiveness in significantly reducing the complexity and resources required to test, minimising time-scales, minimising costs.

Please see here for an example Application Load Testing project report.

Please call us now on +44 (0) 870 2405519

or, for more information, email us at info@mydiagnosys.com