Internet traffic surges by 25% for second year in a row, average US download speed is...

[midian182]

In brief: For the second year in a row, global internet traffic grew 25% over the last 12 months, highlighting how the rate of growth isn't slowing down as more businesses, services, and people rely on being connected. It's one of several facts to come from Cloudflare's 2023 Year in Review report on internet trends and patterns.

Cloudflare's summary of findings for 2023 shows that nearly half of web requests used HTTP/2, with 20% using HTTP/3. Only a third of IPv6-capable requests worldwide were made over IPv6. In India, however, that share reached 70%

Over 40% of global traffic comes from mobile devices, though it's over 50% in more than 80 countries. Desktop was the preferred method of access in the US, with mobile devices accounting for just 33.3% of internet traffic.

Android was the most popular OS for getting online, with over two-thirds of mobile traffic coming from devices running Google's software. Android had an over 90% share of mobile device traffic in over 25 countries and regions while peak iOS mobile device traffic share was 66%. In the US, the split was 46.6% Android and 53.4% iOS.

Another interesting finding is that in the United States, the most common download speed is 0 - 5Mbps, while upload speed is 10Mbps - 15 Mbps. But the average download speed is 201 Mbps and the average latency is 39ms.

Download/upload speeds in the US

A third of global bot traffic comes from the United States, and over 11% of global bot traffic originates from Amazon Web Services. Finance was the most attacked industry, and deceptive links and extortion attempts were two of the most common types of threats found in malicious email messages

Google was once again the most popular general internet service. It was followed by Facebook, Apple, and TikTok. Mark Zuckerberg's social media giant will be pleased to have placed above rival TikTok, which was the most popular service in 2021.

It should come as no surprise to learn that OpenAI was the most popular service in the generative AI category; the company's Wikipedia entry was the website's most visited article of the year. We also saw "AI" crowned as word of the year by Collins Dictionary.

Roblox topped the metaverse and gaming services category for the year, ahead of Xbox Live, Epic Games/Fortnite, and PlayStation. Oculus was fifth, ahead of sixth-placed Steam, though Valve's service was more popular than Oculus from July to September.

Elsewhere, Stripe knocked PayPal off the top spot in the services category. Binance held on to its crown as the most popular cryptocurrency service, Amazon led in the E-commerce section, YouTube was top for streaming, and WhatsApp the most popular messaging service.

Cloudflare writes that the rankings of service popularity are based on analysis of anonymized query data of traffic to its 1.1.1.1 public DNS resolver from millions of users around the world.

Permalink to story.

https://www.techspot.com/news/101203-cloudflare-internet-traffic-surges-25-second-year-row.html

 

[Uncle Al]

Considering the number of new computer users and surges in on-line shopping, etc. I'm not at all surprised in this increase .....

 

[ScottSoapbox]

Considering the number of new computer users and surges in on-line shopping, etc. I'm not at all surprised in this increase .....

New computer users?

Do you mean the recently born?

 

[RandomWAN]

I don't see how the US average is so low: Every ISP is offering more speed at less cost to compete with home 5G alternatives. This is very apparent here where I live where they are now offering home fiber options when historically the local providers tried to avoid doing so. I have to assume they're stacking this with much more rural options/sources of data since anyone near a large population center should have pretty good speed options available.

 

[ScottSoapbox]

I don't see how the US average is so low: Every ISP is offering more speed at less cost to compete with home 5G alternatives. This is very apparent here where I live where they are now offering home fiber options when historically the local providers tried to avoid doing so. I have to assume they're stacking this with much more rural options/sources of data since anyone near a large population center should have pretty good speed options available.

I'm not sure what you mean by "stacking" with rural data, but rural internet speed is terrible in the US. And 5G alternatives are likewise terrible in rural areas.

That's what these stats mean: "the most common download speed is 0 - 5Mbps... But the average download speed is 201 Mbps"

MANY people have 5Mbps or less, but people in cities have speeds far above 200Mbps which raises the average.

The government has repeatedly spent a bunch of money to "improve rural internet" (and is in the process of doing such again), but somehow the money keeps going to not fixing the problem (e.g., providers spent the money upgrading urban internet, went to politically connected startups that then pivoted to something else, etc.)

 

[RandomWAN]

I'm not sure what you mean by "stacking" with rural data, but rural internet speed is terrible in the US. And 5G alternatives are likewise terrible in rural areas.

That's what these stats mean: "the most common download speed is 0 - 5Mbps... But the average download speed is 201 Mbps"

MANY people have 5Mbps or less, but people in cities have speeds far above 200Mbps which raises the average.

The government has repeatedly spent a bunch of money to "improve rural internet" (and is in the process of doing such again), but somehow the money keeps going to not fixing the problem (e.g., providers spent the money upgrading urban internet, went to politically connected startups that then pivoted to something else, etc.)

As in artificially depressing the average based on how you slice the data. Anything near larger towns or bordering cities has significantly high speed available. According to the US Census Bureau, only 15% of the population lives in rural areas. The majority of the rest live near some larger population center. With the 5G infrastructure push, being in range of a cell tower improves your options and combining that with the proliferation of 'doesn't suck' satellite internet, the number should be on an upward trajectory.

I know a number of people who living in rural areas that are still able to game and function online with what is available without using 5G or Starlink. I also wonder if their data gathering takes into account if people are actively searching out better connections or if they just have it for occasional use (where you wouldn't care about the speeds).

 

[ScottSoapbox]

As in artificially depressing the average based on how you slice the data. Anything near larger towns or bordering cities has significantly high speed available. According to the US Census Bureau, only 15% of the population lives in rural areas. The majority of the rest live near some larger population center. With the 5G infrastructure push, being in range of a cell tower improves your options and combining that with the proliferation of 'doesn't suck' satellite internet, the number should be on an upward trajectory.

I know a number of people who living in rural areas that are still able to game and function online with what is available without using 5G or Starlink. I also wonder if their data gathering takes into account if people are actively searching out better connections or if they just have it for occasional use (where you wouldn't care about the speeds).

There is no slicing of data going on. I'm sorry the statistics don't line up with your anecdotal evidence.

And being "able to game and function online" is a pretty low bar.

Being "near" a city is a vague distance. For most things being a reasonable commute is fine: driving to a job or stores. Running fiber in the ground for 30-90 minutes worth of driving distance is a lot of money to commit for a return that drops off with every mile.

Further, the most common download speed is not a doctored-data average - it is merely the mode (I.e., the most common number). If 15% of the people are in the 0-5 speed range and the next highest group is 14% of people at 10,000-15,000 the former is still the mode and the statement is still accurate and true. Understanding the math behind statistics is vital in interpreting this stuff.

Finally, the other groups contributing to low-speed internet are poor people and old people (that don't use it or care), but rural folks don't even have the option available to them.

 

[RandomWAN]

There is no slicing of data going on. I'm sorry the statistics don't line up with your anecdotal evidence.

And being "able to game and function online" is a pretty low bar.

Being "near" a city is a vague distance. For most things being a reasonable commute is fine: driving to a job or stores. Running fiber in the ground for 30-90 minutes worth of driving distance is a lot of money to commit for a return that drops off with every mile.

Further, the most common download speed is not a doctored-data average - it is merely the mode (I.e., the most common number). If 15% of the people are in the 0-5 speed range and the next highest group is 14% of people at 10,000-15,000 the former is still the mode and the statement is still accurate and true. Understanding the math behind statistics is vital in interpreting this stuff.

Finally, the other groups contributing to low-speed internet are poor people and old people (that don't use it or care), but rural folks don't even have the option available to them.

I'm sorry you don't understand. Game and function online is not a low bar, you need low latency and good up/down speeds to be able play competitive games; you won't have an enjoyable time if you're shooting at ghosts. Gaming requires reasonable download speeds more and more as most gaming content is delivered online and the average game size ~50GB now with many of the bigger AAA titles reaching 100GB+. With multi-gigabyte uploads, having a 200Mbps or less connection means you'll be waiting a while.

If you want mileage quotes on how close someone is to a city, I could go dig that up if I felt so inclined, but reasonably is means you live somewhere that doesn't require a large travel time to reach a city (greater than 30 minutes).

"Another interesting finding is that in the United States, the most common download speed is 0 - 5Mbps, while upload speed is 10Mbps - 15 Mbps. But the average download speed is 201 Mbps and the average latency is 39ms."

The average matters if you're using it in arguments to redefine what qualifies as broadband in the USA (since there's been a push to redefine it, any changes to it filter down to us who pay for access). If the ISPs are pressed to roll out 100 Mbps or greater to more remote areas, the customer is going to be forced to eat those costs and they won't be able to recoup it from those they're forced to supply.

According to the article, the referenced document places 0-5 Mbps as the most common download speed (mode), but does it reference how they came to that number? Is it their home speeds or are they lumping in what people are connecting with while on a mobile device away from home? Given that the referenced information points to 40% of global traffic comes from mobile devices, how big is that percentage in the USA? How are they collecting this mobile traffic data? Is it just considering the source OS or is it accounting for cellular vs being on wifi?

The phrase "lies, damn lies, and statistics" is true and I was simply musing with the mild hope that either the Techspot article author or someone who had read the entire Cloudflare article might have gleaned that information and would share since I haven't had the time to read through it in depth since I've seen a surge in speeds available while the costs associated with those speeds are falling. Next time I'll fully flesh out my statement in detail as to not have you take umbrage.

 

[ScottSoapbox]

I'm sorry you don't understand. Game and function online is not a low bar, you need low latency and good up/down speeds to be able play competitive games; you won't have an enjoyable time if you're shooting at ghosts. Gaming requires reasonable download speeds more and more as most gaming content is delivered online and the average game size ~50GB now with many of the bigger AAA titles reaching 100GB+. With multi-gigabyte uploads, having a 200Mbps or less connection means you'll be waiting a while.

If you want mileage quotes on how close someone is to a city, I could go dig that up if I felt so inclined, but reasonably is means you live somewhere that doesn't require a large travel time to reach a city (greater than 30 minutes).

"Another interesting finding is that in the United States, the most common download speed is 0 - 5Mbps, while upload speed is 10Mbps - 15 Mbps. But the average download speed is 201 Mbps and the average latency is 39ms."

The average matters if you're using it in arguments to redefine what qualifies as broadband in the USA (since there's been a push to redefine it, any changes to it filter down to us who pay for access). If the ISPs are pressed to roll out 100 Mbps or greater to more remote areas, the customer is going to be forced to eat those costs and they won't be able to recoup it from those they're forced to supply.

According to the article, the referenced document places 0-5 Mbps as the most common download speed (mode), but does it reference how they came to that number? Is it their home speeds or are they lumping in what people are connecting with while on a mobile device away from home? Given that the referenced information points to 40% of global traffic comes from mobile devices, how big is that percentage in the USA? How are they collecting this mobile traffic data? Is it just considering the source OS or is it accounting for cellular vs being on wifi?

The phrase "lies, damn lies, and statistics" is true and I was simply musing with the mild hope that either the Techspot article author or someone who had read the entire Cloudflare article might have gleaned that information and would share since I haven't had the time to read through it in depth since I've seen a surge in speeds available while the costs associated with those speeds are falling. Next time I'll fully flesh out my statement in detail as to not have you take umbrage.

I'm sorry you don't understand the anecdote about your rural friend finding gaming acceptable, followed by your definition of what acceptable gaming must be, is not the compelling point that you think it is. Neither is pointing out that large games would take a long time to download on a slower connection.

"lies, damn lies, and statistics" is a funny quote but only "true" for those that don't understand statistics. However, I understand that not everyone spends a good portion of their work studying data and interpreting the statistical output of research studies, so I'm just going to move on.

 

[human7]

According to the article, the referenced document places 0-5 Mbps as the most common download speed (mode), but does it reference how they came to that number? Is it their home speeds or are they lumping in what people are connecting with while on a mobile device away from home? Given that the referenced information points to 40% of global traffic comes from mobile devices, how big is that percentage in the USA? How are they collecting this mobile traffic data? Is it just considering the source OS or is it accounting for cellular vs being on wifi?

From the CloudFare document:
"By aggregating the results of speed.cloudflare.com tests taken during 2023, we can get a geographic perspective on connection quality metrics including average download and upload speeds, and average idle and loaded latencies, as well as the distribution of the measurements."

The article also breaks down ISP usage among other things, so the answer to your question of why the average is around 200 Mbps is probably in there somewhere. I would point out that the average is a statistical metric that is vulnerable to influence from outliers, such as 1 Gig speeds and higher. The median would be far lower, though not necessarily in the 0-5Mbps range.

There is no slicing of data going on. I'm sorry the statistics don't line up with your anecdotal evidence.

To be fair to RandomWAN, ISPs have been known to inflate their speed metrics quite a bit. This study comes from a (hopefully) more reputable source, however we don't know how representative the results from `speed.cloudfare.com` are compared to other speed tests. Even so, the statistics seem quite reasonable to me.

 

[lripplinger]

"Average speed" in the U.S. means nothing. In the valley I live, I have 200/200 fiber optic. But then my parents who live 15 minutes from me, only have garbage DSL (I think they are lucky to get 20/5 tops). This is all the same local ISP.