Will Covid-19 Traffic Kill the Internet?

This is the question being asked all across the industry as the volume of data traffic has leaped upward due to students and employees working from their homes. We got our first glimpse of the impact of the crisis when Verizon announced a week into the crisis that they were seeing a 22% increase in data traffic in their network. More recently AT&T announced a 27% increase in network traffic. In perhaps a peek at what might be coming, Italy, which has been in a lockdown for longer than the US has seen a 90% increase in Internet traffic.

The answer to the question differs depending on somebody’s perspective of the network. For example, Evan Swartztrauber, described as an advisor to the FCC, says that the US Internet network is handling the surge in traffic just fine. He says the increased volume is significant, but it’s not at the same level as what is seen during the Superbowl or the finale of Game of Thrones. That’s reassuring news to hear, but he’s talking from the perspective of the big Internet POPs and the long-haul networks that carry Internet traffic from city to city. Even his answer is a bit glib because we’ve just seen more than a year’s growth in traffic in a matter of weeks and there must be places in the Internet backbone that need to be beefed up to meet the increased demand.

The question that matters is if Internet performance is getting worse for the average user, which is a question about the last mile network. I’ve been checking in on clients to understand the impact. So far, everybody with a fiber-to-the-home network says they are weathering the increased volumes, although several clients are looking into increasing bandwidth in a few parts of the network, such as between the core and field huts. Several clients who operate HFC or DSL networks have told me that their biggest problem is with upload speeds. People working from home as well as students are using a lot more upload bandwidth as they communicate with office and school servers. Gamers also need significant upload bandwidth. These technologies were not designed to handle significant amounts of uploaded bandwidth and customer performance is seriously degrading.

Many clients also say that they are increasing the bandwidth needed to connect to the Internet. Luckily most of them can do this easily, but some rural clients are constrained on the ability to easily add more bandwidth.

What nobody is talking about is the last-mile networks that were already broken. For example, I helped a rural county to get citizens to take speed tests right before the pandemic and we found almost no rural households in that county with broadband speed greater than 5 Mbps – and most are far under that modest number. These customers are served with DSL or fixed wireless broadband, and the local telco and WISPs are obviously bandwidth restricted either due to older technology or due to lack of backbone bandwidth.

Rural networks that are already underperforming might easily collapse under increased bandwidth usage. A 30% increase in usage won’t cut speeds by just 30%, the extra usage is likely to crash the networks. A large portion of rural America already has dreadful broadband. There are terrible ramifications if a network that is only delivering 3 Mbps broadband today gets further stressed. Degraded usage means that a home where a student might have been able to connect to a school server before Covid-19 might now be unable to maintain a connection. Good luck to somebody trying to connect to an office server as they work from home for the first time. And considering that some of these stressed rural networks have upload speeds measured in kilobits per second, good luck to anybody wanting to make a video connection for school or working from home.

Perhaps it’s true that the overall US Internet is not in danger yet of collapsing. Networks are going to see additional stress if the shelter-at-home restrictions carry through April and into May or June. What all of the national headlines are missing is that many rural Internet networks were barely functional before the pandemic. I’ve talked to numerous rural businesses in the last year that don’t even have adequate broadband to sustain a credit card transaction. I hear from homes across the country where the Internet is too slow, or latency too high to sustain connections to a school network to do homework. The current burst in new traffic is going to mean that the Internet performance for many rural users is going to go from barely functional to non-functional.

We might see a little relief if some of the biggest bandwidth users of the web cut back on broadband demand. Google announced that they are going to reduce the quality of video signals from YouTube as a way to cut back on the volume of data hitting networks. There is pressure on Netflix to do the same. AT&T announced that Netflix’s traffic volumes have hit an all-time high. Netflix announced that it is going to reduce traffic volumes by 25% in Europe but hasn’t made the same claim yet for the US. Unfortunately, these fixes are unlikely to make a big difference. the problems in last mile networks is due to having many more Internet users than before the pandemic, and the sheer number of users along with their attempts at using bandwidth-hungry applications is going to kick rural networks in the teeth.

This pandemic has highlighted the horrendous inadequacies of rural broadband. The shortfalls of rural broadband already existed, but with the added traffic volumes, rural broadband is going to significantly worsen. Unfortunately, we didn’t see much funding to help rural broadband as part of the recent stimulus plan. I’m pretty sure politicians with rural constituents are going to hear a lot about this – at least constituents with enough bandwidth to tell their story.

One thought on “Will Covid-19 Traffic Kill the Internet?

  1. To your point about when the stress will break the network…VA looks a lot like NC in terms of rural. Our Governor just extended our shelter at home to a June 10.

    Sent from my iPad

    >

Leave a Reply