Akamai Diversity

The Akamai Blog

Situational Performance in Q1 2015

In June 2013, Akamai announced the latest release of Ion. Ion is designed to meet the unique challenges of optimizing the desktop and mobile Web experience. One feature of Ion is a capability known as Real User Monitoring (RUM). RUM takes performance measurements from real Web users to provide developers with insights into performance across a multitude of devices and networks. Ideally, RUM is used in tandem with synthetic testing to generate a comprehensive picture of a user's Web experience to help developers best calibrate their applications.

Real User Monitoring (RUM) Methodologies

There are two key RUM measurement methodologies that allow data collection. First, is the "navigation timing" API, often referred to as "navtiming." The Navtiming API allows JavaScript to collect page load time component information directly through the browser.

The second collection method is through a framework for Web page timing, such as Web Episodes. Of the two methodologies, navtiming is preferred. Having said that, the navtiming methodology is not supported for every user agent. In particular, Apple's Safari browser on iOS does not support navtiming data collection as of the first quarter of 2015. Beyond Apple, Google added support for the API on Android 4.0 Ice Cream Sandwich and onward, while Microsoft has supported it since version 9 of Internet Explorer. For more information on data collection methodologies, head to theThird Quarter Situational Performance blog post.

The underlying data in this section was collected with navtiming; therefore, as noted above, it does not include measurements from users of Safari on iOS devices or older versions of Android, Internet Explorer, or Safari on OS X. The countries included were selected based on several criteria, including the accessibility of measurements from users on networks identified as mobile and those identified as broadband. Furthermore, countries need to have more than 90,000 measurements from mobile networks during the first data collection period to qualify.

Average Page Load Time Over Broadband Connections

After reviewing the average page load time measurements for broadband connections across the first quarter of 2015, Turkey, South Korea, and Bolivia had the fastest page load times. Turkey led the pack with a remarkable 857 ms load time, while South Korea and Bolivia trailed with speedy 1.6 second average load times. As was the case in the fourth quarter of 2014, Brazil had the slowest average load time at 6.1 seconds. Sri Lanka had Singapore rounded out the bottom three with an average load times of 5.0 seconds.

Average Page Load Time Over Mobile Connections

When it comes to mobile networks, Turkey, Bolivia and Indonesia posted surprisingly fast load times at 602 ms, 670 ms, and 689 ms, respectively. All three countries topped last quarter's leading 850 ms mobile load time. Conversely, Singapore, Taiwan, and Canada had the slowest average mobile page load times at 10.7 seconds, 9.6 seconds, and 8.9 seconds, respectively. The increase in page load times over the fourth quarter suggests that more content-rich pages are being downloaded on mobile networks.

Mobile Penalty

The report also compares average broadband page load times to those on mobile connections. When the two connection types are compared, we can determine the relative "mobile penalty." The mobile penalty indicates how much slower a page loads on an average mobile connection versus a broadband connection. As noted previously, the ratio suggests only one factor in the overall user experience--average page weight. Therefore, it should not be taken as a raw comparison of the mobile versus broadband network speeds.

Across the 46 countries surveyed in the first quarter, the mobile penalty ranged from 0.3x in Indonesia to 3.0x in Canada--a broader variance than was observed in the previous quarter. In total, 21 countries had a mobile penalty lower than 1.0x, meaning that average page load times were faster over mobile connections than on broadband connections. The lowest mobile penalties were observed in Indonesia, Bolivia, and Thailand, all of which had average mobile page load times that were less than half their average broadband page load times. Singapore, Hong Kong, and Taiwan joined Canada with the highest mobile penalties. All four countries had pages load twice as fast, on average, over a broadband connection than on a mobile connection.

More Information

Going forward, as more customers integrate the Real User Monitoring capabilities and more platforms offer support for the navigation timing API, the scope of the Situational Performance section may be expanded. For the entire picture, download the Q1 2015 State of the Internet Report.