Surprising Solutions to the Mobile Content Delivery Dilemma

From EMEA to APAC to the EU, growth in mobile devices and consumption of Internet bandwidth have grown at an incredible rate. One of the features of mobile devices that consumers like best is the ability to have their experience tailored to their geographic location, demographic profile, behaviors and preferences. These capabilities have greatly enhanced the user experience by providing users with information that is most relevant to them, whether that means finding the closest gas station, locating the best sushi restaurant within a city block or finding the best deal on a new pair of shoes while riding the bus to work.

Understandably, global mobile data usage is growing at a CAGR of 53 percent and will reach more than 30.6 exabytes per month in 2016, according to Cisco's Visual Networking Index. Interestingly, the Middle East and Africa region has the highest CAGR at 117 percent, followed by Asia Pacific (83 percent), Latin America (73 percent) and Central and Eastern Europe (71 percent). North America (55 percent) and Western Europe (52 percent) sit at the bottom of the pack in terms of growth.

This is the age of mobile, and it's had a profound impact on business. Organizations that embrace mobile and truly understand the value of user experience have outperformed their peers - think of Jim Cramer's FANG stocks (Facebook, Amazon, Netflix and Google) - and have dominated their respective markets.

However, mobile has to be done right for businesses to experience its benefits. It introduces a host of challenges and complexities that, if not properly addressed, can hinder the user experience, resulting in customer churn, shorter sessions and lost revenue. Mobile content delivery is highly dependent on devices, operation systems, infrastructure and network capabilities - meaning there are literally billions of combinations, all changing in real time, that have to be accounted for in order to meet consumer expectations. This is especially true in the last mile - the segment of the route from the cell tower to the device.

Mobile Performance Issues
Mobile application performance has found a friend in intelligent infrastructure solutions, which help application developers ensure their apps - and their business - don't fall behind. This is important in all regions of the world but is extremely important in emerging markets, which are also the fastest-growing markets, where bandwidth is limited and infrastructure is not as robust as more developed parts of the world.

There are some excellent examples of geo-specific challenges that impact mobile application performance in Twin Prime's State of Global Networks Report (Q1, 2016). In this report, billions of network requests were analyzed from over 150 countries, 3,600 operators, 12,000 device models and more than 100 content types to gain a deep understanding of how networks perform.

There's more interesting information in the report, including good examples of how network (3G, HSAP, 4G, Wi-Fi) performance issues impact mobile application performance, broken down by city, country and region. The report is definitely worth a look if you want to get a better understanding of the factors that impact network performance and work to build more performant mobile applications.

DNS: The Solution to Mobile Content Delivery Challenges
Any time a person types "ILoveShoes.com" or taps an app on their mobile phone, the first thing their device does is ask the internet, "Which IP address should I go to?" The internet responds with an IP address. At its most basic level, that is how DNS works. But in a world of mobile devices, multiple data centers, CDNs, fiber cuts, traffic congestion and servers spinning up or going down, what appears to be a simple answer becomes much more complicated and important.

To assist with this complicated issue and arrive at the optimum answer, intelligent DNS solutions - which use application, network and infrastructure data - can be deployed to optimize mobile content delivery and avoid global internet potholes based on geographic information and dynamic telemetry. Some use cases for intelligent DNS include:

• Telemetry: It's not always the geographically closest data center that's the optimal place to send users. Network congestion, fiber cuts, server health and load distribution are chief among factors that negatively impact mobile content delivery and are challenges that intelligent DNS solutions can solve. Advanced traffic routing solutions can extend these capabilities to include real user monitoring (RUM) data as well as AppDex data from virtually any source.
• Geo: Uses geo-location, including country, state/province or latitude/longitude, to route users to the geographically closest data center or, where needed, geo-fence users to a predefined set of end points (made even more accurate with EDNS-client-subnet support).

Since the earliest days of the internet, DNS has been quietly doing its job behind the scenes. As the digital ecosystem has shifted from the PC to mobile, the old ways of managing DNS simply stopped being effective. Fortunately, intelligent DNS offers application developers a way to make traffic routing decisions much closer to the end user and improve user experience at a very granular level.

Better Performance via Machine Learning
Part of this intelligence comes from being able to make predictions based on data, and that's where machine learning comes in. Machine learning has been around for a long time but, because of the tremendous computing power that is required to ingest, interpret and accurately predict, machine learning was limited to those with access to high-powered computers - typically found at a major university or government agency.

Today, thanks to the universal accessibility that the cloud enables, developers across the globe can gain access to the compute power necessary to use machine learning in a broad set of applications. This is good news because solving last-mile challenges involves understanding literally billions of user, device and network data inputs and combinations, making machine learning a critical tool for solving mobile content delivery challenges.

The image below reveals the advantages that machine learning-accelerated users have over non-accelerated users. In this example, 88 percent of the app's global traffic saw an increase in speed of 75 percent. In addition, users who needed 400ms to 500ms to download content were able to download that content in 150ms or less. Overall, accelerated users benefited from a significant performance advantage and, based on a great deal of user experience data, those users are more likely to improve metrics like session length, session interval, retention rate and abandonment which, in turn, has been shown to boost revenue and LTV.

Mobile content accelerators that are based on machine learning can make a positive impact on both the client-side and server-side aspect of mobile content delivery, including:

• Selecting the best origin server for content retrieval
• Reducing packet loss during transport
• Optimizing cache to increase your app's cache hit ratio
• Selecting the best protocol and protocol parameters based on context
• Minimizing DNS latency and optimizing routing
• Defining the optimal congestion control and TCP ramp-up based on user context

The world has gone mobile and there's no going back. Old methods of content delivery fall short, so modern mobile content accelerator solutions were created to fill in the gaps. They are intended to work alongside CDNs and supplement commonly used application performance management and monitoring tools. They are a low-cost-for-performance solution when deployed using SDKs. As data growth continues, such a configuration will equip your organization for the present and prepare it for what's to come.