To detect and act on issues before end user quality of experience (QoE) is affected, service providers need a new breed of traffic classification capable of fine-grained analysis at 250ms-intervals
When Verizon launched its test-man commercial in the early 2000s, voice quality over mobile took center stage. But today, voice is just one component of an increasingly complex universe of capabilities that include video, chat, messaging, mapping, AI, immersive AR/VR, the IoT, and so much more.
Even before COVID, Gen Z was going to push service providers toward vastly greater and more unpredictable demands for capacity and performance. COVID just accelerated that evolution. Now, previously siloed experiences for working, collaborating, learning, entertaining, and socializing have blurred into one collective experience, and the conformity to “9-to-5 or “peak hour” constructs has been obliterated.
This has triggered a massive reboot of business systems and communications, with people, enterprises, and institutions learning to interact and engage within different boundaries and rules.
To remain relevant, or better yet, to truly “innovate,” service providers will have to embrace that it’s not the “network” they are selling, but rather “app QoE.” They will have to rapidly build out new services and revenue streams around the apps that are most important to different groups and subgroups of customers. This will require an understanding of not only macro trends across millions of subscribers, but also micro trends across smaller subsets of users. Only then can operators proactively detect congestion and improve network performance for optimal QoE, as well as move toward monetizing what’s important about QoE for their consumer and enterprise customers.
Growing Importance of Frequent, Granular Traffic Classification
To understand how apps are affecting networks, traffic has to be classified and stratified according to which apps are generating the most traffic, and over which devices, when, and where. This type of insight requires new levels of visibility, and should come through the lenses of:
The data should be fine-grained through measurements and correlations of multiple data points that inform different functional areas like: network and capacity planning, operations, big data, market development, revenue generation/monetization, and customer care.
The granularity of the data will depend on its speed and frequency. Sandvine’s Application and Network Intelligence provides insight at 250-millisecond intervals so that operators can see what’s really happening with users and application traffic across different services, content types, locations, devices, and intentions. At smaller intervals, it’s possible to see the smallest issues, even before they are perceivable to end users, as shown in the image below.
In this traffic classification, there is a visualization of what’s happening during a video conference call over a popular enterprise platform. In red are quality drop-offs, the longest being eight seconds, during which listeners might be saying “we can’t hear you.” After a pause, the network and application try to catch up by speeding up the audio, which then sounds rushed to listeners. The smaller dips in green (the 2nd and 5th in particular) illustrate where data is picking up on other issues the human ear is not perceiving. If those dips are frequent, they can be an indicator of problems to come.
Having insight at smaller intervals give service providers early indicators of problems, such as network degradations or changes during times of high or low bandwidth usage. Smaller intervals also help predict what may happen with bit rates as people talk, interact, share, and engage, driving proactive resolution before QoE is compromised.
The ability to visualize app performance at millisecond granularity will be far more effective than what was possible with traditional network analysis, which maxed out at 5-second intervals. That is no longer sufficient for today’s apps and services.
Equally important to the shorter intervals are flexibility and agility, which is why we update our application classification logic on a weekly basis, ensuring customers and partners keep up with trends as they come and go.
Why It Matters?
Having deep and expedient insight translates into fewer surprises, as well as optimal QoE, greater revenues, and lower costs. As an example, our Global Internet Phenomena Report had a section “A Terabyte Isn’t What it Used to Be” that was based on genericized traffic classifications of customers’ networks, revealing a sharp uptick in 1 TB-per-month households as well as contributing factors, such as: surge in video, gaming, and work-from-home apps; increases in QUIC and encryption; pandemic-driven downstream and upstream phenomena.
By digging into the specific categories of apps, and then the specific apps within those categories, we could see the correlation between certain events and changes in subscriber behaviors and usage patterns. We could see the impacts or cascading impacts not only on networks, but on other apps and other customers’ experiences.
This type of insight will only grow in importance for our customers, especially as more “households” become extensions of enterprises. The goal is to provide a seamless experience over different devices and access types, which is why we increasingly support service providers using a cloud-based core for both fixed and 5G networks. As this evolution happens, operators will want to know how consumer and enterprise apps are affecting their networks, and conversely, how their networks are affecting those different categories of apps.
Just think of what happens today in an “average” home, and how many different experiences under one roof can affect a service provider’s brand. Each experience can open a window to new services for each person: a parent using videoconferencing and chat, while a spouse is streaming a favorite show, simultaneous to a teen gaming with friends, or a sibling taking a virtual class.
Knowing whether the remote worker successfully collaborated over Microsoft Teams, or whether the Roblox gamer enjoy his shared experience with friends, or whether Squid Game successfully streamed, will help the service provider continuously roll out new services based on the experiences they deliver around these and other popular apps.
The same is true in the Enterprise and industrial IoT scenarios. Businesses of all types will need help keeping up with rapid and frequent changes to their business models and business processes. Through edge computing and 5G network slicing, operators will have a chance to deliver positive experiences for even the most performance-intensive apps, both on prem and in the Cloud. Operators will also help improve internal comms and collaboration, drive automation, and collect and analyze data for new revenue generation.
At the heart of all these opportunities will be data, analysis, visualization, and insight about the ways in which gamers, streamers, remote workers, and the IoT impact underlying networks, and hence, app QoE.
To learn more about how Sandvine can deliver insight to your CTO organization, Operations, Network Planning and Engineering, Market Development, and Customer Care, contact us and check out our Application and Network Intelligence Portfolio, 5G Service Intelligence Engine, and Analytics.