Posted on 12 April 2018

Often in IT projects, the last item to be considered is a performance test. This is a mistake! Poor performance leads to unhappy users. The key to any go-live event is to ensure that your system or application is ready and able to perform well under load (i.e. under the pressure of multiple users). By performance testing with LoadRunner and the help of an experienced performance engineer, you can ensure that your system or application—no matter how simple or complex—is ready to go-live.

Let’s set up a scenario.

You’re undertaking a large project with high pressure expectations and a multitude of stakeholders— and unfortunately, you’re over budget and behind schedule. Do you still schedule in performance testing instead of going straight to production?

What about if you have an internal company system that is slowing down, thus now slowing down your staff from completing work. Considering the bad press that comes along with an application or system that crashes under load, is it worth taking the risk of trying an upgrade?

In both of these scenarios, the answer is yes—performance testing always de-risks projects, making them more likely to succeed and less likely to negatively impact your users.

How does LoadRunner work?

Using LoadRunner, we can emulate user traffic from a user’s perspective. It allows us to create load and volume using the business processes that users perform and thus gives us the ability to see how the system will cope. It’s capability to emulate user traffic and correlate associated metrics (system resources, web servers, network, etc.) is extremely powerful. It’s as close as you can get to a crystal ball in IT—letting you predict what will happen within your system before it goes to prod and affects your users.

LoadRunner comes with a wide range of protocols that it can emulate such as WEB/HTML, RDP, Citrix, SAPGUI, and more. Using its scripting tool, LoadRunner Virtual User Generator (VuGen), performance engineers are able to record and script the different business processes so they accurately represent what users action. Then, in the LoadRunner Controller, a scenario is created so it models the anticipated load on a system. This is used to execute the test and collate results. Finally, in the LoadRunner Analysis tool, we can go over the results of the test and correlate metrics together to view the various aspects of the test. We can also work with the data to discover trends within the results that will hopefully pinpoint any potential performance issues.

Take the example below captured from LoadRunner’s analysis tool after a test:

LoadRunner has a long history in IT, with the first version released in 1989. It's an incredibly mature tool, with a solid reputation.

This was a test conducted for a large data-heavy web-based system. Notice the high response times across the board for each of the different processes—can you imagine the reaction from users of this system if response times are going over 100 seconds for each request?

Another example is testing a new version of an application. The expectation is that the new version should just work…right? A performance test later, looking at the below graph, it would appear that one of the transactions had a performance defect introduced as it now has constant spiking response times. Users now have to wait for an extended period of time to view documents—something that wasn’t an issue before and functional testing did not find. Only while executing and analysing a performance test was this noticed.

 

In 2000, a simplified version of LoadRunner called Astra LoadTest was launched. Today, if you kick LoadRunner hard enough you might even see an error dialog referring to Astra LoadTest.

So how do you go about solving the above problems? Firstly, ensure you get performance testing done to begin with! Then comes the analysis of the data you have collected:

  • Are resources high?
  • Has code been modified?
  • Is the database under-performing?
  • Is there a particular business process that contains high response times?
  • Is it a common component of the application that struggles?

Sometimes the answer is easy and obvious; other times it requires in-depth technical understanding of the environment as well as the results gathered by LoadRunner. Being able to combine performance test execution with in-depth technical analysis is crucial to meaningful results being found.

JDS consultants have more than a decade of experience working with LoadRunner, and we are one of the only Premier Partners of Micro Focus in Australia. This gives us an edge when it comes to performance testing that is unmatched by other performance testers. If you are in the process of introducing a new system or application, make sure you schedule a performance test with JDS. Contact our Micro Focus team today on 1300 780 432 or at microfocus@jds.net.au.

Already have a system or application in place, but looking to make it faster? Take advantage of our One Second Faster solution—five days for less than $5k to get a performance health check.

Our team on the case

Brad Halkett

Length of Time at JDS

Since January 2015

Skills

  • Highly experienced performance tester
  • Application performance monitoring
  • Strong technical skills across a range of technologies

Workplace Solutions

  • AppDynamics
  • HP LoadRunner
  • HP SiteScope
  • HP Virtual User Generator
  • JMeter

Every day, do something that people want.

Nick Wilton

Consultant

Length of Time at JDS

8.5 years

Skills

Primary: Software security, Performance optimisation

Secondary: DevOps, Software development, Technical sales

Workplace Solutions

I help clients to solve problems like:
  • Is my application secure?
  • How do I manage threats?
  • Will my application perform when I need it to?

Workplace Passion

It’s all about managing risk whilst driving business confidence in technology and software solutions. That’s what I’m passionate about.

Leave a Reply