Manage Network Congestion, Not Applications
Network congestion is a major cause of support calls to ISPs, particularly from frustrated customers complaining that their internet is slow.
As the number of applications and internet-connected devices in the home continues to rise, regional providers need up-to-date tools to address and manage network congestion in an effective and sustainable manner.
Doing so will make for happier customers who are less likely to churn and more likely to recommend your service to others. It will also help keep your ISP competitive by lowering support costs and limiting time-consuming truck rolls.
Manual traffic throttling and application-aware packet inspection are unnecessary and ineffective methods that are out of date with an internet that’s rapidly evolving.
Managing network congestion proactively using capacity planning and set-it-and-forget-it traffic management is the way forward for ISPs who care about providing their subscribers with the best possible experience on a constantly changing internet.
What is Network Congestion and What Causes It?
Network congestion refers to when subscriber quality of experience (QoE) is negatively impacted by high latency caused by saturated links and poor queue management.
This can be caused by subscriber “self-congestion,” i.e. when a customer is using their full plan bandwidth. An example of this would be a large game or OS update saturating the subscriber’s connection. Streaming video services can also negatively impact the latency-sensitive experience of gaming and VoIP applications. This is the source of those “Netflix is slow” or “My Zoom call keeps cutting out” support calls with which most ISPs are by now very familiar.
Side note: If you’re interested in learning how Netflix actually works (and how it’s not really a stream at all), check out our blog and webinar on the subject.
The reason that the customer experience degrades rapidly as links approach congestion is that most devices employ simple “first-in, first-out” (FIFO) buffering schemes which add latency and eventually cause packet loss across all flows.
A congested network might also be the result of hardware issues, such as overloaded access points. Indeed, we’ve found that APs are the most common congestion points in any network. An added complexity for operators trying to troubleshoot underperforming APs is that different models will behave differently under load, requiring specialized knowledge and additional time to diagnose and fix the issue.
Deep Packet Inspection is Not the Answer
A once-popular but now outdated method of attempting to manage network congestion is Deep Packet Inspection (DPI). This traffic management approach monitors packets travelling through the network using a mix of complex techniques like heuristics, signatures, and machine learning. It then determines, where possible, which application generated each packet. The idea is that packets from high-use applications (e.g. Netflix) can be given higher priority, and so improve the experience of the subscribers using that app.
Though this was a useful method in the internet’s early days, there are many issues with continuing to use DPI on modern networks. Here are just a few examples:
- Applications on the internet are constantly changing. New applications are developed and existing ones change all the time. Keeping up-to-date is a constant game of cat and mouse, which is only getting more challenging over time.
- New standards are continuing to be developed, which makes it more difficult to accurately identify applications (e.g. TLS ECH)
- A network that relies on prioritizing specific applications is brittle. What if the subscriber behavior changes to use different applications or if the traffic pattern changes so that certain applications are no longer identified properly?
- Who gets to decide which applications are high priority and which are low? Some might designate gaming updates as “low priority,” for example, but try telling that to the estimated 3 billion+ gamers worldwide! It’s likely that you have just as many gamers on your network as you do Netflix watchers, so why prioritize one over the other?
Managing your network and trying to deliver the best possible experience to your subscribers shouldn’t be a time-consuming, manual game of whack-a-mole, where regional operators are expected to constantly make value judgements on which traffic is and isn’t “important.”
How to Reduce and Manage Network Congestion
We believe it’s better to solve network congestion problems before the customer calls in to complain, rather than try and identify which applications were responsible after the fact. That’s why Preseem uses FQ-CoDel, an active queue management (AQM) technique used to dynamically size queues and optimize for superior latency and throughput, even under heavy usage.
With FQ-CoDel, the classification function is much simpler and is based on the behavior of the individual network flow. If the flow has a lot of packets in a short interval, it’s treated as bulk. If the flow has a few packets in a short interval, it’s treated as interactive. Applications like voice and games, which have a small but constant bitrate, automatically fall into the latter category.
As a result, you don’t have to spend time and energy shaping for individual applications—instead, you set it and forget it, and all of your subscribers get a good experience no matter what application they’re using.
That said, however, traffic management is not QoE in and of itself. It’s a great tool you can use to mitigate congestion when it occurs, while still ensuring fairness across subscribers on the congested resource (e.g. an access point) to maintain a good experience.
However, the best way to actually manage and reduce network congestion is (drum roll please) to avoid it in the first place. To which you might reply, “Thanks, Captain Obvious” or something slightly saltier. However, what we mean is that proactive capacity planning is your best defense long-term against network congestion issues.
According to Cisco’s annual report, the number of internet users is growing at a rate of 6% annually, while the number of devices in use within the home is growing even faster, at 10% year over year. Bandwidth needs also continue to grow, with a 28% jump in 2022 alone. This means planning for the future is essential to make sure you have the capacity to handle the increase in subscribers and devices that are inevitably coming your way.