July 2018 Web Server Survey

In the July 2018 survey we received responses from 1,663,673,364 sites, 220,573,018 unique domains, and 7,734,417 web-facing computers. This reflects a gain of 33.4 million sites, 2.80 million domains, and 120k web-facing computers.

In the web-facing computer market, nginx experienced the largest increase (+3.24%), contributing more than half of the overall growth and surpassing two million web-facing computers this month. As a result, it was the only major vendor to experience a gain in market share. nginx also experienced the largest growth in terms of active sites (+7.87%), and top million sites (+0.71%), further increasing its market share in those metrics. Although nginx has slightly fewer domains than last month, it is generally seeing long-term growth in all metrics, and has gained around 10 million domains over each of the past three years.

Apache remains ahead in terms of domains, active sites, web-facing computers, and sites within the top million. However, it continues to see both long and short-term losses in many metrics, with steadily decreasing counts of domains (-0.53%), active sites (-1.15%), and top million sites (-0.98%).

Microsoft's server software leads by total number of sites, of which it had 670 million this month (a 40.28% share), but sits behind Apache and nginx in most other metrics. Microsoft retains second place in total number of domains with a 26.02% market share, and experienced the largest increase this month of nearly 1.4 million domains. This has increased the gap between it and third-placed nginx, which has been steadily gaining on Microsoft, now leaving them 3.24 percentage points apart.

The one metric in which all major vendors continue to see ubiquitous long-term growth is web-facing computers. Since July 2017, the total number of web-facing computers has increased by 1.14 million (+17.3%). This rate of growth is more than double that seen in the year before between July 2016 and July 2017. Although there have been overall increases in the number of computers running Apache and Microsoft webservers, both have continued to lose market share to their faster growing competitor, nginx.

Google's .app TLD takes off

.app is one of many new gTLDs from ICANN's New gTLD Program. ICANN appointed control of the .app TLD to Google after it won the rights in a $25 million auction against twelve other applicants in February 2015. Google hopes that the appropriately named .app TLD will appeal to websites providing applications and application-related services.

The .app TLD only opened for public registration fairly recently on the 8th May 2018. This month's survey found 185k unique domains running websites under the .app TLD. Whilst many of these domains return parking pages, a number of fully-fledged sites, such as cash.app, have already been launched, with the Netcraft survey detecting over 15k active sites.

One of .app's marketed features is its inclusion within browsers' HTTP Strict Transport Security (HSTS) preload lists. The .app TLD is one of only thirteen where an HSTS preload policy has been applied to a whole TLD. This TLD-wide HSTS policy means that modern browsers will only access .app websites over a secure HTTPS connection, protecting users from trivial man-in-the-middle attacks.

Total number of websites

Web server market share

DeveloperJune 2018PercentJuly 2018PercentChange
Microsoft601,258,60536.88%670,209,36440.28%3.41
Apache361,165,37822.15%366,023,11922.00%-0.15
nginx336,529,38220.64%287,045,45017.25%-3.39
Google22,946,3071.41%23,347,2901.40%-0.00
Continue reading

Most Reliable Hosting Company Sites in June 2018

Rank Performance Graph OS Outage
hh:mm:ss
Failed
Req%
DNS Connect First
byte
Total
1 www.choopa.com Linux 0:00:00 0.000 0.312 0.009 0.237 0.238
2 Bigstep Linux 0:00:00 0.004 0.251 0.074 0.150 0.150
3 Hyve Managed Hosting Linux 0:00:00 0.004 0.185 0.077 0.155 0.155
4 CWCS Linux 0:00:00 0.004 0.300 0.085 0.163 0.163
5 www.dinahosting.com Linux 0:00:00 0.004 0.309 0.090 0.182 0.182
6 GoDaddy.com Inc Linux 0:00:00 0.009 0.528 0.005 0.020 0.020
7 www.viawest.com Linux 0:00:00 0.009 0.488 0.009 0.266 0.266
8 EveryCity SmartOS 0:00:00 0.009 0.239 0.072 0.378 0.378
9 Swishmail FreeBSD 0:00:00 0.009 0.255 0.086 0.171 0.172
10 One.com Linux 0:00:00 0.013 0.359 0.058 0.175 0.175

See full table

The most reliable hosting company website in June 2018 belonged to choopa.com, it was the only site to respond to all of Netcraft's requests. Choopa.com offers cloud hosting, dedicated hosting and colocation in its own primary facility in Piscataway, New Jersey as well as smaller facilities in Los Angeles, Amsterdam, and Tokyo. The website has achieved an uptime of 100% over the past 21 months, and a 99.98% uptime since Netcraft began monitoring the website in December 2009.

The next four hosting company websites only failed to respond to a single request each during June. The lower average connect time for BigStep's website, taking an average of 74ms, grants it second place above Hyve Managed Hosting, CWCS and dinahosting.com. Bigstep offers "bare metal" cloud hosting from datacentres in the UK, USA and Romania. The service claims to provide the flexibility of cloud hosting but without the associated overhead and performance reductions of virtualization.

Third placed Hyve Managed Hosting's website has appeared in the top ten 11 times in the past 12 months, the most of any monitored website. Netcraft has never observed an outage for www.hyve.com, giving it a 100% uptime record since Netcraft began monitoring it in June 2016.

Eight of the top ten most reliable hosting company websites used the Linux operating system in June. FreeBSD and SmartOS both make a single appearance to complete the top ten.

Netcraft measures and makes available the response times of around thirty leading hosting providers' sites. The performance measurements are made at fifteen minute intervals from separate points around the internet, and averages are calculated over the immediately preceding 24 hour period.

From a customer's point of view, the percentage of failed requests is more pertinent than outages on hosting companies' own sites, as this gives a pointer to reliability of routing, and this is why we choose to rank our table by fewest failed requests, rather than shortest periods of outage. In the event the number of failed requests are equal then sites are ranked by average connection times.

Information on the measurement process and current measurements is available.