Nobody enjoys waiting for content to load. At Twin Prime, we analyze millions of requests daily and study traffic patterns in order to eliminate the infamous spinning wheel. Through our GLAS (Global Location based Acceleration Strategies) database, we’ve processed billions of network requests globally, which gives us unique insights about network performance. Understanding which factors impact performance is key to dealing with mobile performance issues, and our data exemplifies how variability determines mobile performance.
In this inaugural report, we analyzed half a billion requests across 100 countries, 200 carriers, and 1000 different types of devices to examine three key factors that contribute to variability: Network, Device, and App.
1. WiFi Speed Myth
Many people will tell you that they prefer WiFi over cellular networks because WiFi is more reliable and faster than cellular; however, our data shows that this is not the case. Below, we compare performance of LTE and WiFi on iPhone 6 devices in several cities. The devices are running the same version of iOS and downloading 200-250KB images in the same app.
Above: In all of these cities, WiFi takes longer to download 200-250 KB files.
The following graph highlights the difference in download times. WiFi is significantly worse than LTE where the bars go farther right. When discussing speed in terms of milliseconds, it is important to understand that people tend to notice a 200ms difference, and multiples of 200 are especially noticeable.
As you can see, modern cellular networks are often noticeably faster than WiFi. According to the data for the last few cities on the graph, when WiFi is faster, it is often only slightly faster.
2. Devices, Devices, Devices
There are tens of thousands of types of mobile devices, and performance varies among them. Even comparing download speeds on iPhone 6 devices using the same carrier (AT&T LTE) in the same city (Philadelphia), running the same app and downloading similar file sizes reveals a significant speed difference. On average, iOS 8.4 devices downloaded 200-250KB images at 5.861 megabits/second, and iOS 8.3 devices downloaded images at 3.661 megabits/second. That’s a 37.5% speed difference simply due to using a slightly different version of the device! Even though these devices are downloading the same image sizes in the same location, they experience a very different range of download speeds. Furthermore, there is a large interquartile range even on the same OS version:
While this is not enough to draw a broader conclusion about iOS versions, it does highlight the considerable variability in users’ experiences. Consequently, there cannot be a one-size fits all strategy to improve performance across all these thousands of types of devices.
We can also look at a single device type and examine how performance changes by city. For example, our data shows that the same version of iOS running on iPhone 6 devices on the same LTE carrier running the same app downloads 200-250KB files faster in Los Angeles than in Houston and Philadelphia.
3. What about the App?
Finally, performance varies within an app depending on the type of content. Different strategies are necessary to optimize loading large files compared to small files. Most people assume that smaller files download faster, but the situation is more complex. Due to the variability in mobile, there are many cases where there is little difference in perceived download times across a wide range of file sizes. For example, the following download time data is from iPhone 6 devices on iOS 8.3 and AT&T LTE in Philadelphia:
Performance also varies depending on the type of app. Here, we compare download speed between a popular social media app and a popular commerce app. Commerce apps tend to be bulkier; they fetch data from multiple domains, so it takes longer to display all of the content. Social media apps, on the other hand, likely fetch data from a single administrative domain.
Putting it all together
The popular perceptions are that you should optimize your app for locations with poor network performance, and that simple optimizations within the app can make it work better everywhere. The reality is a lot more complex due to the incredible amount of variability in mobile. We’ve shown that:
- LTE is often faster than WiFi
- Performance varies by device, location, and content
- Our intuitions about how file sizes affect download times turn out to be inaccurate.
- Don’t compromise on quality; optimizing performance is more complicated than sacrificing rich content
The above insights help explain why optimizing mobile app performance is complicated; variables such as the user’s device and location are completely outside of your control.
By recognizing the variability in mobile and adapting to it, you can boost your app’s performance. Twin Prime accomplishes this with GLAS technology. GLAS boosts your app’s reliability by detecting and analyzing all of these variables, then choosing the right strategies to optimize your app’s performance.
Source: Satish Raghunath
Via: Twin Prime
Image: Pocket Now