Mobile Performance Testing: Demystified
A look at how to approach performance / load testing extranet mobile applications
By: Rebecca Clinard
Oct. 12, 2012 11:00 AM
How does the approach to performance / load testing extranet mobile applications change from testing other RIAs? In some ways, not much. Yes, the user activity for every mobile application will be unique, but the same goes for every web application. Different technology stacks, different deployments, different load patterns, different types of users profiles, and different content delivery… All make each web application, in its own right, unique. The approach to methodical performance testing for capacity planning and identifying scalability issues remains the same.
The mobile application could be a web site accessed via a URL or it could be a native mobile application loaded onto your device. The list of mobile devices keeps growing, but currently the most popular are Tablets, PDA’s, Droids, Blackberry, iPhone, iPad, Smartphones, etc. Supporting testing from these devices is important, but what’s even more important in your mobile load testing strategy is to generate traffic that represents the way users are actually going to access this mobile application. Whether it’s an ecommerce application being accessed via a Safari browser or an installed mobile app which makes webservice calls out to a feed, the load testing tool must generate the correct type of user connections and generate realistic load. This is the exact same challenge with testing non-mobile web applications as well.
Mobile applications are being developed and deployed rapidly, using a variety of cutting edge protocols and behaviors. The load tool must support the latest technologies in order to effectively address mobile apps. You need the ability to record a script either straight from the mobile device, from an emulator, or from a mobile browser. Handling behaviors such as asynchronous behaviors of PUSH technology becomes a requirement in many mobile applications. As important as the design of a performance test case scenario, now the load generation piece must either mimic the unique characteristics of mobile devices or have the ability to generate the load straight from the device. For example: headers, number of connection threads, etc. All these characteristics are important elements to emulate when conducting web and mobile performance testing.
What’s different about mobile? Mobile applications are being built using network efficiencies which aim in relying less on the intermittent characteristics of mobile networks. It is the variable network conditions that cause delays in response times which in turn affect the duration that ports or sockets are kept open — an environmental resource usage that is frequently seen with mobile applications. It’s because of this variable network connectivity that the user experience isn’t always an absolute known: the speed of the network depends on your device, your network plan (WIFI, 3G, 4G, ATT, etc), your geographic location, network connectivity, the network usage vs. bandwidth, and so on. Your load tool needs to have the option of emulating these bandwidth speeds to more accurately capture response times. Developers concentrate on what can be controlled: building efficient mobile apps which require less network overhead. There are many techniques to reduce the network traffic, all of which contribute to making responses less dependent on the network. “Conservation” is the approach of reducing network roundtrips by decreasing embedded requests, using local storage on the device for caching static files, enabling transfer compression, avoiding redirects, minimizing data content size, reducing number and length of cookies, removing lint from code (white spaces and comments), organizing the delivery for incremental rendering, aggregating requests and using PUSH behaviors. Creating lighter-weight mobile applications allows the overall end user experience to be less dependent on the device network vulnerabilities.
In the end however, the approach to load testing an extranet mobile application remains the same. You have goals: number of concurrent users, expected response times, load patterns etc. Using the right load tool, you can emulate this activity. You then identify capacity and saturation points and alleviate these bottlenecks in order to scale to a higher workload. You already know the drill for doing this and if you don’t, please refer to 11 Tips to Becoming a Better Performance Engineer. It sounds simple, but mobile performance testing is an interesting and vast topic so check back in for future posts on the subject from some of my colleagues.
Latest Cloud Developer Stories
Subscribe to the World's Most Powerful Newsletters
Subscribe to Our Rss Feeds & Get Your SYS-CON News Live!
SYS-CON Featured Whitepapers
Most Read This Week