Guest post by Noam Green, VP of Marketing at Flash Networks
Mobile video traffic continues to increase at a breakneck pace, pushing the limits of mobile networks. Last year mobile video traffic was up more than 80% compared with 2012, reaching 1.5 exabytes per month at the end of 2013, according to Cisco. Mobile video growth is expected to continue at this tremendous growth rate, growing to 69% in 2018, where over two-thirds of the world’s mobile data traffic will be video.
The introduction and deployment of LTE networks offers real promise for handling the increase in traffic. However, subscribers’ insatiable appetite for mobile data, including high definition video, is increasing faster than 4G network capacity.
According to a study on US LTE networks, 20% of mobile Internet traffic is Netflix video, compared with only 5% on 3G networks, and overall 50% more traffic travels on 4G networks compared with 3G networks. Part of the increase in mobile video traffic is also fueled by the video capabilities of Vine, Instagram, and Facebook.
Mobile video traffic, if not managed correctly, can negatively impact all network users, creating a ripple effect in the network which may resemble a similar concept known as the butterfly effect. In chaos theory, the butterfly effect is where a small change at one place in a deterministic nonlinear system can result in large differences to a later state. For mobile subscribers, video traffic in a congested network creates a butterfly effect by degrading other types of traffic and reducing overall quality of experience even before it affects video traffic. Video traffic consumes large blocks of bandwidth due to a larger packet size and clogs the network proportionally more than lighter types of network traffic such as images and text. As a result, mobile video traffic poses a threat to user experience for all subscribers, including those who are not viewing video on their mobile devices.
Mobile Video, the Eighteen Wheeler Truck
Mobile video consumes a disproportionate amount of network resources, compared to its contribution to network traffic. How much bandwidth does it take? Here is a specific example. An average LTE site has a total throughput of 40 Mbps, and one HD 1080p You Tube Video requires 8 Mbps. This means that if 6 users are simultaneously viewing HD video on a single cell, the cell is already pushed beyond capacity.
Here is the problem: in addition to resulting in buffering or jittery video for those subscribers running YouTube on mobile devices, all subscribers are affected. Video packets are transmitted in larger buckets of TCP packets (also referred to as having a larger TCP window size), meaning that in a congested network they may get higher priority in the TCP flow than other types of traffic which send over small buckets of data. Due to the way TCP is designed, this inherent unfairness creates situations where smaller volume traffic, such as browsing, get slowed down due to the larger video packets flooding the network.
When you consider the comparative size of TCP packets on the network, text is like a compact car, images are like cars, and video is like an eighteen wheeler truck. When there are large amounts of video traffic on the network, it is like trying to merge with rush hour traffic driving a car, when everyone else on the highway is driving a huge rig. It is virtually impossible. As a result, when there are large amounts of video traffic that create congestion, the same amount of network bandwidth is only able to manage a significantly smaller number of transactions.
This is happening at the same time that another trend is gobbling up more and more bandwidth. The mobile Internet is being pushed beyond capacity also by the sheer size of web pages that are being downloaded.
Strangeloop revealed that between 1995 and 2010 the average web page size grew from 14KB to 498 KB. By the end of 2014, however, Web Performance Today expects the size of the average web page to exceed 2,300 KB, which is a 44% growth compared to 2013. But while web pages have grown in size, they have also grown in the amount of objects they encompass, from an average of 2.3 objects in 1995 to over 83 objects in 2012.
Despite the fact that mobile video is hogging bandwidth and web pages are heavier, subscribers expect the same download speeds they experience on desktops. This is concerning for operators since long load times can lead to web stress and increase customer churn. A study by Radware revealed that network delays of as little as 500 milliseconds can result in up to a 26% increase in peak frustration. In fact, according to Accenture’s Mobile Web Watch, mobile Internet connection speed is important for 97% of users, and many of them are willing to pay for faster connections.
Reducing the Butterfly Effect
In order to increase the number of transactions transferred per second, video optimisation can be applied to reduce buffering and free up space for less bandwidth-intensive transactions, such as web browsing. In addition to smoother video viewing, video optimization reduces the volume of video traffic, thus freeing up space for shorter transactions to fill the gaps. More transactions lead to faster browsing speeds, and overall higher quality of experience. Since speed is increasingly the most important differentiator for mobile operators, mobile video optimization provides the needed boost to all traffic for an improved user experience.
Mobile video optimisation which implements real-time bit-rate adaptation provides both on-line and off-line transcoding for superior performance and applies data reduction techniques only when there is a high level of congestion. Video quality also can be monitored to ensure that any data reduction that is applied does not have a negative impact on image quality.
Perhaps, even more importantly, video optimization can work in tandem with TCP optimisation which accelerates mobile networks by 30% in addition to efficiencies gained through specific mobile video optimisation techniques.
Retaining subscribers in this competitive environment is challenging. Operators that want to minimize the butterfly effect have mobile optimisation options which, in addition to improving download speeds, video quality, and the smoothness of video streaming, can improve the quality of user experience for everyone.