Mobile devices are super useful thanks to apps. For mobile app development, latency must be managed to continue the mobile convenience.
Let's take a quick look at a growing performance metric that may not be the first on your dashboard needs, but can be a needed metric for those managing the next great app. That metric is latency.
Latency is the measure of a time delay in a system. It's meant to measure packets switched within a given network. That may sound esoteric and cryptic, but latency has a value as an indicator of network capacity.
Latency is calculated as part of page speed, where it represents delay:
Speed performance = Bandwidth (bits/sec) X Latency (sec)
Latency usually occurs at a higher rate on a mobile device than on a desktop - an O'Reilly video regarding latency notes about 3-4 times difference between mobile and desktop on average.
Latency has been labelled as the new bottleneck for expanding the customer experience on mobile, because for an app or website elements, it can be a factor for loading delays. In layman terms, it means the lower limit of bandwidth capacity needed to move information from server to client - a value for those apps that share data frequently in their design (which is a fancy way of saying "most apps"!). So the larger the latency, the more possibility of delay.
So how is latency best measured? Latency can be measured within a web analytics solution. Google Analytics monitors latency as several metrics within its Page Speed reports using a combination of the Page Timing and User Timing reports. Google details how the reports work in regard to latency in this technical post.
But care should be taken into how the metrics are gathered. Google, for example, notes that it samples its page speed data. Such sampled metrics may not consistently represent how a webpage is rendered, so that can be a challenge in having an exact value. Page speed measurement tools can offer better accuracy (These are covered in this Zimana post). Consider also a packet sniffer such as Charles or Fiddler. These tools can simulate a call to a hosting server based on a limited amount of traffic. Such a call can reveal problems for mobile and tablet devices operating with a limited signal and indicate what can be done to improve app performance.
