Pingdom recently performed three interesting hosting-related studies that are also interesting generally regarding global Internet behavior. All of them relate to where in the world websites are hosting their sites – what different nations and cities are being used the most – by looking at the GPS coordinates of the servers (alongside other information).
One is a general study of the popularity of certain countries for website hosting. A second breaks down that information into specific cities. The third and final study is an exploration, via analysis of websites using country-specific top-level domains (TLDs), of how many sites within certain countries use servers that are “onshore” – meaning internal to their own nations.
Finally, I will look at how ghost servers haunt the Internet: acting creepy; moaning; shouting “boo” at unsuspecting website visitors; and in the process, scaring many children and elderly web users out of their chairs.
Where in the World is the Web? – General Study Methodology
We obviously think of the Web as “global” (it is, after all, the Worldwide Web). However, it’s not evenly distributed throughout the world, as we all know. Looking at where the servers are located that are hosting websites is one way of understanding how the Internet is spread out across the planet.
Of course, these studies do not look at how many servers are used by each site or company, but it at least shows us the behavior of different sites related to geographical location of their servers. Also, tiny sites were not included: all sites that were scanned were in the Alexa-Netcraft top 1 million sites at the time the studies were conducted. Plus, for the third study, no .com or other gTLDs (generic, ie non-country specific) were included because it related the location of the site’s ownership with the location of its servers – impossible to perform with the generic TLDs.
In other words, the studies do not give us a perfect or complete picture, but the stats gives us a general idea of physical location of web hosting. This tells us where the Internet – at least as hardware being accessed for page loads by companies and individuals – is located throughout the world, in some cases down to the city. I’ll discuss methodology a bit more in each study’s section (below).
Ghost servers, by the way, do not have a defined methodology. They tend to have certain characteristics though. They are wisps of surreality, populating sites that only exist for a moment. The ghost servers use the sites to freak everyone out and then recede into the abyss from whence they sprang (both the site and the hardware do this in unison, as the server moans a soft yet resonant howl of pain and longing).
Study 1: USA, Server Nation
The first study I will look at from Pingdom focuses on the countries. The study reviewed the Alexa top 1 million and clustered those sites into the top 100 nations where their servers are located (a follow-up to a 2012 study conducted by the company). Bear in mind, these figures are not in any way gauged by population, geographical size of the country, or any other factors. Quantity of people and scope of land obviously helps boost the US’s numbers over many other nations.
The information for all the studies – the raw data – was all gathered on February 27, 2013. It was attained using a script developed by Pingdom that allowed it to GPS-scan all the Alexa 1 million in rapid succession. For the first two studies, 907,625 sites were scanned. Out of the remaining sites, 52,539 could not be scanned, and 39,836 did not have any observable GPS coordinates. Since the GPS coordinates were irrelevant to the 3rd study, the scanned number of sites was 947,461 instead (although, as described below, many of those sites had to be removed because they were generic TLDs).
What about the ghost servers, you ask? No ghost servers were scanned for this study. Ghost servers and the sites represented by them are not in the Alexa 1 million because they cannot be tracked; and, furthermore, if you attempt to study them, they start screaming. Ghost server science, therefore, is considered inhumane if it is at all in-depth. I even feel kind of bad about writing what I do here, fearing that somewhere a ghost server might be wailing in agony because of my actions. Sorry, spirits.
Worldwide … Well, Sorta
191 nations serve as a home, hosting-wise, to at least one of the Alexa 1 million websites. According to Worldatlas.com, different sources list the number of countries on the planet at just below 200 as well – between 189 and 196. Just looking at it in terms of inclusion of one “top” site, then, the Web is distributed worldwide. However, the numbers for the primary countries are massive. The top ten countries are as follows:
- United States: 421,228
- Germany: 70,587
- China: 35,908
- United Kingdom: 35,500
- Russia: 35,254
- France: 34,498
- Japan: 29,898
- Netherlands: 25,632
- Canada: 18,116
- Poland: 12,109
Compared to 2012, the numbers are similar at the top. The US lost a tiny bit of ground in the first position, dropping from 43% to 42% of all hosting for the top websites. Germany, China, and the UK are all in the same positions as a year ago. The US is still the home-base for the Web though, in this sense. The country hosts 6 times #2 Germany’s figure, the latter accounting for 7% of the global total.
France and Japan lost a small bit of relevance as Russia moved up two slots from its previous 7th-place spot. Poland has also increased its standing – in 2012, it was #13 on the list.
Ghost servers are primarily located in Liechtenstein. Many people think that the ghost servers are trying to convey recipes, herbal remedies, and other pieces of cultural information in whatever language is spoken in Liechtenstein. The only problem is that no one knows for sure what language is spoken in the country. Everyone is pretty sure it’s either French or German … possibly its own language, if that exists. Maybe one of us should look this up.
Comparison Grouping by Continent
Let’s now take a brief look at how the United States relates to the two most prominent continents, Asia and Europe. The US, as noted previously, hosts 421,228 or a 42.1% share of the world’s highest-traffic sites. Somewhat amazingly, Europe grouped as a whole is still more than 10 percent below the United States: 314,317 sites, representing 31.4% of the sites. Asia is far below Europe at 11.5%, with a total of 114,571 of the Alexa 1 million. The remaining 15.0% are located in other continents and non-US North American locations.
In 2012, Pingdom pondered in their coverage of their initial study on the same subject whether the US was going to be surpassed by other global locations. Nothing much has changed since last year as far as that goes. However, as this year’s Pingdom analysis notes, since Asia is at 25% of global Internet users, it will be continually interesting to see whether or not their hosting industry starts to make pace with their population of users.
Ghost servers are not like typical hardware. Rather, they are believed to be constructed out of a mixture of cobwebs, dreams, fog, and eerie music (the last of which, in its physical form, looks exactly like it sounds, whatever that means).
Study 2: Houston, Server City
Now let’s take a look at the second Pingdom study: city analysis. 7,936 cities around the world host the Alexa top 1 million sites. Like the above figures, though that number seems reasonably well-disbursed, the top-ranked cities account for a large chunk of the action. 223,206 of the top 1 million – which is roughly equivalent to 22.3% — are all hosted in just 10 cities. What are the top cities for hosting? Take a look:
- Houston, Texas – 50,598
- Mountain View, California – 29,594
- Dallas, Texas – 24,822
- Scottsdale, Arizona – 23,210
- Provo, Utah – 20,691
- Ashburn, Virginia – 14,871
- San Francisco, California – 13,214
- Chicago, Illinois – 13,125
- Beijing, China – 11,273
- New York, New York – 10,006
Again, we see a major disparity even between the top two sites and a 5-fold difference between the 1st and 10th cities on the list. In fact, the top three cities – Houston, Mountain View, and Dallas – account for 10.5% of the hosting of the total 1 million sites! We also see the major and continuing impact of the United States on the size of the Web, with the US making up 9 of the 10 top positions.
Ghost servers should not be taken lightly. When you see a ghost server in person, always approach it cautiously, and never under the influence of alcohol or drugs. Observation of ghost servers should be conducted in the same manner as if you were operating heavy machinery. Approach clear-headedly; functionally; and wearing heavy work gloves and a hardhat.
Study 3: South Korea, Onshore Central
Of course the United States has many sites hosted within its own borders, but what countries are the most “loyal” to their home country regarding their hosting? Keep in mind, these figures are not perfect by any means: none of the gTLDs (generic top-level domains), including .com, could be scanned. This study simply looked at which countries had the most onshore hosts for their sites based only on the country code top-level domains (ccTLDs), to determine what percentage of those sites stay within the country for hosting.
What are the top countries? And what are the respective numbers of sites in the Alexa top 1 million? Here they are, and note that they are listed in order of percentage of sites, not quantity, with the quantity in parentheses. Below, you’ll see a shorter list of the nations with the most ccTLDs hosted onshore.
- South Korea – 97% (1,750)
- Vietnam – 93% (2,260)
- Germany – 92% (25,469)
- Japan – 91% (14,188)
- Czech Republic – 90% (4,736)
- Lithuania – 88% (1,051)
- Bulgaria – 87% (825)
- Thailand – 85% (699)
- Kyrgyzstan – 84% (102)
- Hungary – 84% (2,619)
Here, also, are the top five countries with onshore-hosted ccTLDs – again, from the Alexa top 1 million:
- Russia – 43,002 (.ru)
- Germany – 25,469 (.de)
- United Kingdom – 17,558 (.uk)
- Brazil – 16,991 (.br)
- Poland – 14,235 (.pl)
Pingdom notes as well that this figure is only representative of the ccTLDs. Over 509,000 of the sites in the Alexa top million use the .com TLD, just as one example of the limitations inherent to this study.
Ghost servers smell like gravy and taste like asparagus. Strange, right? Don’t eat them though: bad, bad gas, according to one blog writer who regrets the experience and will go unnamed so no one thinks he’ll eat anything that comes his way.
Have any thoughts on these studies? Any ideas related to the numbers, or anything that perhaps looks surprising to you? Please continue the conversation below if you like. Thanks for reading. Also, um, no one can ever prove that I ate a ghost server. I … did not do that.