Startup Load Impact is today announcing a new cloud-based load testing service specifically designed for mobile apps. The load — or stress — testing service can simulate traffic loads to a mobile apps’ API and analyze performance factors so that developers looking to enable a mobile first strategy via API can get a real picture of their end users’ potential experiences. Founder and CEO Ragnar Lönn spoke with ProgrammableWeb on the eve of the new service announcement.
“When it comes to installable mobile apps, they often access their server-side functionality through some kind of API,” said Founder and CEO of Load Impact, Ragnar Lönn. “So oftentimes, load testing a mobile app means that you are, in effect, load testing an API. You simulate the activities of the app — the HTTP requests it generates to its API servers — and see how the servers hold up when 10,000 app users call your API functions at the same time.”
Simulating real-world mobile APIs
One of the key features of the new mobile load performance testing service is that it lets users create a range of scenarios based on actual hardware and network configurations most common amongst their user base.
By first reviewing a combination of their mobile and API analytics, developers can set simulation levels that match expected mobile patterns as they scale: performance can be tested with a certain proportion of API calls being made over a 3G network, with another proportion set to to test performance over an LTE network. Similar simulations can be set to carve up API calls made depending on the mobile user’s operating systems.
Explaining the new features, Lönn said:
“When we simulate a mobile client — whether it is a mobile user running a mobile web browser and accessing a standard web site, or it is a mobile user playing the Candy Crush Saga app [that makes API calls to their server to manage game play], we can generate the same kind of traffic for the servers to handle, as real users.
If the average mobile user has a network connection speed of, say, 384 kbit/s (old generation 3G) we will not let our simulated client load data faster than that from the servers. In the past, every simulated client/user in a load test loaded things at maximum possible speed, at all times. This will, of course, result in a very skewed test result, that might tell you your app can handle a maximum of 1,000 concurrent users, while in reality you could handle a lot more (or less).
Apart from simulating network connection speed, we also simulate network latency, which is as equally important for performance as connection speed. And just like connection speed, it also affects how ‘heavy' a client is for the servers to handle.”
Trends in mobile first performance
While the new mobile load performance testing will be useful for businesses preparing a mobile-first strategy, or for apps and API calls that expect a high fluctuation in sudden usage — such as around a localized event — for the main part, the service is expected to help the full range of Load Impact customers to extend their suite of testing tools to the mobile environment. To date, it has been a lesser acknowledged area requiring performance testing.
“Only about half of the companies with mobile apps test their mobile code”, revealed Lönn, who says there are some understandable reasons for this. “I think the reason so many companies aren't doing load testing, mobile or otherwise, is that it has always been seen as expensive and difficult. People developing mobile apps are no exception - they also try to avoid things that they think might be difficult, expensive and time-consuming.”
Load performance testing is another example of the barriers to entry that cloud-based services are smashing, with Load Impact and other alternatives like Loader.io and GTmetrix allowing businesses to not only perform load tests, but to stream performance analytics via an API directly into their broader analytics and monitoring systems.
Mobile is one area of performance testing that API developers supporting business’ mobile first strategies cannot ignore. At DeveloperWeek earlier this year, Jay Srinivasan from Appurify showed that 83% of users would abandon an app after two buggy crashes, while a study by Shunra Software found 48% of mobile users are less likely to use an app after it fails. At the recent API Strategy and Practice Conference, the real-time data panel shared that up to around 80% of new app solutions are based on some sort of real-time data need, meaning performance of API calls over mobile will become even more crucial than they are already, especially when those API calls made over mobile have implications for keeping track of the latest updates made to a business’ underlying databases.
API developers can trial Load Impact’s new mobile app testing service for free on the Load Impact cloud service site.
By Mark Boyd. Mark is a freelance writer focusing on how we use technology to connect and interact. He writes regularly about API business models, open data, smart cities, Quantified Self and e-commerce. He can be contacted via email, on Twitter, or on Google+.