In a previous post, I’ve discussed how load testing with the cloud can enable your team to conduct more efficient and more realistic large scale tests while saving you (time and) money. With this all in mind, how do we go about choosing the correct cloud load testing solution?
There are many cloud load testing solutions that will enable you to make use of the cloud in some way. However, very few enable you to capitalize on the opportunities that load testing with the cloud offers. As I’ve discussed, load testing with the cloud offers clear advantages over traditional load testing in certain circumstances, but the tools you use are even more important to the quality of your tests.
When considering a cloud testing solution, ask the following questions:
- To what extent does the solution integrate with the cloud?
- Will the solution enable us to conduct realistic tests?
- Does the solution support unified tests inside and outside the firewall?
- Is the solution easy to use, or will we spend weeks learning and configuring it?
- Does the solution include full-featured reporting and decision-making modules to help our team make the most of the results?
- Does the solution support the technologies we used to build the application?
Much of this is outlined in the Neotys Cloud Whitepaper, but for the purposes of this blog, let’s briefly talk about each of these:
Integration with the Cloud Platform
If you choose a solution that is not integrated with one or more cloud platforms, there are several manual steps you’ll need to perform on your own. First, you’ll need to learn how each platform you’ll be using works, including its limitations and constraints. Second, you’ll need to build, maintain and access the machines you wish to use.
Load testing solutions that offer integration with the cloud simplify and accelerate the steps needed to use the cloud infrastructure. With the proper solution, connecting to, and utilizing the cloud should be possible with just a few mouse clicks.
Realistic Load Tests
Simply moving to the cloud is not always enough to ensure the most realistic tests. Real users often have access to less bandwidth than a load generator in a cloud data center. With a slower connection, the real user will have to wait longer than the load generator to download all the data needed for a web page or application. This has some major implications as your performance results won’t be fully reflective of what your real users will experience (here we go again with the inaccurate load testing).
When choosing a load testing solution, look for one that provides a bandwidth simulation feature that limits bandwidth to ensure that the virtual users download the content of the web application at a realistic rate. This capability is particularly important when testing mobile applications, because mobile devices typically operate with less bandwidth than laptops and desktops.
Unified Lab Testing and Cloud Testing
While including the cloud in our load testing is obviously a huge advantage, it does not mean that we should exclude testing from within our own firewall. Both types of testing should be included – if we are only doing one or the other, we are again running into a disadvantage. Likewise if we use different solutions for these different tests.
When choosing a load testing solution, you should make sure that it supports lab testing so you can confirm the performance on your application(s) internally, before you introduce additional variables by testing over the internet. This allows you to shake out any early performance issues with smaller loads before you execute tests from the cloud. Finding these issues earlier in the application’s lifecycle will lower costs associated with not only the cloud but also those attributed to “late stage bugs”.
Making sure that you use a single solution for both “internal” (within the lab and/or firewall) and “external” (via The Cloud) testing allows your engineers to reuse scripts across both types of tests. If you choose different solutions, you may have to create a new set of scripts for your external testing, costing you both time and money.
Ease of Use
As every test engineer knows (not just the performance testers, the functional testers can relate too!), testing is almost always performed under tight time constraints. Delays in development or other phases earlier in the lifecycle result in less time for the test engineers to do their jobs. The pressure is on to deliver results as quickly as possible. With this kind of stress, there is no place for a solution that is hard to use. Finding a tool that is easy to use is imperative. Not only does it save time and money, but it will allow you to get more testing done in the time you have – ensuring a higher quality application. While an exhaustive list of “ease of use” features might be best suited for another blog article, some key things you should look for are:
- Easy recording of a virtual user profile (preferably in one click).
- The ability to easily define advanced behaviors (with structures such as conditions and loops) via a graphical interface, without a need for writing code or test scripts.
- Automatic handling of session parameters. This is probably the most challenging and time-consuming part of in the development of load testing user profiles. Find a tool that does this for you automatically.
- Easy results comparisons. Generating graphs and charts based on a lot of raw data can be excruciating – a tool that automates this ability is a must.
Analysis, Monitoring, and Reporting
Creating realistic scenarios and running comprehensive load tests is only the part of the story. The final steps of course are to understand how your application behaved while it was subjected to the test load. Did it meet performance requirements? If not, why? To do this analysis, you need a powerful set of analysis tools to analyze the data generated during the test.
Make sure you choose a solution that allows you to easily analyze the collected data, creating actionable reports that describe the performance of the tested application. I’ve seen too many a test engineer spend hours, if not days taking raw data from a load testing tool and pushing it through Excel or a statistical tool to create graphs, charts and reports. A tool that allows for easy analysis not only makes this process pain-free but can increase the collaboration among the different stakeholders of the application’s infrastructure.
Making sure your load testing solution contains a comprehensive monitoring system is also essential when you need to find the root causes of a problem. If you are only looking at performance measure collected from the end-users point of view, you’re missing half the picture. Being able to use a monitoring system allows you to understand what was happening within the infrastructure of your web application is critical in identifying any root issues. Moreover, the selected load testing solution should allow for easy correlation between data collected from the end-users point of view and data collected from any back-end servers. Without this ability, results analysis becomes extremely cumbersome as well as potentially inaccurate. Too many times in my load testing career I’ve tried to correlate performance data collected from a load testing tool with that of data collected from separate application server and database server monitoring tools. An almost impossible task.
Support for Web Technologies
This one is pretty simple. If your application has been developed using more advanced web technologies, make sure the solution you choose has the proper support. With applications built with Adobe Flex, Microsoft Silverlight, Real-Time Messaging Protocol (RTMP), AJAX push technologies, etc. becoming more and more prevalent, the proper technology support is critical. Without this support, effectively testing the performance of your application is next to impossible. I’m sure that any of you that have tried to load test a Siebel application are nodding your heads as you read this.
While considering the above topics to search for a cloud provider, you may also want to consider using MULTIPLE cloud computing providers. There are several advantages to this approach.
First, multiple providers may allow you to test from more geographical regions. Earlier, we’ve talked about how testing from multiple regions can provide a more realistic test scenario. With this in mind, combine the regions available from multiple providers can lead to an even MORE realistic scenario. In other words, if a single provider does not give you the geographical coverage you need to emulate (based on where the real world users of your application “live”), aggregating multiple regions can be very useful.
Second, if you are executing exceptionally large scale tests, engaging multiple providers simultaneously allows you to bypass any limitations that a single provider may place on bandwidth or the number of machines in use.
Finally, using multiple cloud providers enables you to detect potential network issues at the cloud provider level. Let’s say that during results analysis of particular test run you noting that there is significantly worse performance measure from load generators from a particular cloud provider while all other load generator show acceptable performance. With this data, you can safely conclude that there is a problem (temporary or not) with that provider and not your application. If you had locked yourself into a single cloud provider, you might be limited in your ability to conduct accurate large-scale tests.