Internet / Networking

Why hasn't the internet collapsed?

Dennis Thankachan
Everyone is on the web all the time now. WFH means cloud application utilization and high-def video calls. Socialization also means cloud application utilization and high-def video calls. Households have gone from zero active users during non-peak hours to multiple active users soaking up bandwidth all the time.

But the internet hasn’t collapsed.

In fact, for most, the internet experience of late has been oddly normal. Things are functioning as they should, and users who can are operating from home without major issues.

Should that be the case? In this post, I’ll examine how much network utilization has gone up and what that has meant thus far for broadband networks globally.

How much has ISP network utilization increased?

A lot.

According to the NCTA, peak US cable broadband upload traffic has increased >30% while download traffic has increased nearly 20% since 3/1. It looks like network activity peaked in early-mid April and has been on the decline since.


The NCTA report also notes that wi-fi data traffic and wi-fi calling have increased significantly (vs. mobile) and networks are supporting more connected devices than ever. Additionally, peak utilization hours, normally in the evening, now occur in the afternoon.

Cloud infrastructure and collaboration tools (namely UCaaS) have also seen significant increases in utilization.
Microsoft stated that it has seen a near 8x increase in use of its cloud services in areas enforcing social distancing or shelter-in-place orders. It has also seen significant increases in usage of its Teams collaboration service and Windows Virtual Desktop feature.

Zoom Communications and Cisco’s Webex have seen similar increases in usage. Zoom has even reached 300 million daily meeting participants per day!

So how has this impacted networks?

So far, not much. There are two things to evaluate here - how often ISP networks have seen outages and how much shared bandwidth has declined per user.

On the outage side, not much has occurred. See below for a chart of weekly ISP network outages since early March from the ThousandEyes Internet Report. You can see that while outages increased slightly nearing ~250 / week, they haven’t increased all that much.


See below for a similar chart on cloud service provider outages. Here, you can see a noticeable increase in outages, but that is coming off a low weekly base.


Although the lack of downtime is encouraging, the storyline around bandwidth per user is less encouraging. Everyone knows that more people on a shared bandwidth network means less speed for all, and that’s played out recently as well. Here is some data from Fastly.


Italy and the UK have seen huge decreases in speed. Perhaps this is why the EU has asked Netflix and YouTube to cut the quality of their video to support speeds? In the US, the situation is less dire, but we’ll see if that continues to hold true.
According to the NCTA report cited above, cable networks have not seen overwhelming congestion just yet. This is encouraging!

Why the hell is stuff still working!?


For one, most networks have been provisioned to accommodate for peak utilization bandwidth at all times. It just so happens that users are now at or near peak utilization all the time! This results temporarily in bandwidth reductions for users, but doesn’t “break the internet” per se. As peak bandwidth needs have gone up over the years, ISPs have upgraded electronics and transport to keep pace. That core infrastructure is doing its job well today.

Further, the internet was designed with redundancy and a lack of path dependence in mind. Getting packets to point A to point B can occur by transmitting packets of many different paths, so we’re not actually dependent on any specific circuit operating at a point in time. Further, internet protocol allows for us to know when packets are lost and retransmit data if that occurs. This means true last-mile downtime shouldn’t occur too often at scale.

According to the article cited above, the internet was originally designed in the Cold War with robustness in mind due to fear of a potentially imminent nuclear attack on affecting any individual network site.

We’ll see how long this robustness can last.

If your business is dealing from WFH issues, we can help! Click here to reach out or schedule a meeting, or fill out our questionnaire to inquire about UCaaS or internet solutions that could be helpful to you.

What are you waiting for? Let’s get you pricing.

Not ready to buy? Stay up to date on our product!