|Rank||Company site||OS||Outage hh:mm:ss||Failed Req%||DNS||Connect||First byte||Total|
|1||New York Internet||FreeBSD||0:00:00||0.000||0.161||0.082||0.166||0.494|
|3||www.netcetera.co.uk||Windows Server 2008||0:00:00||0.003||0.109||0.120||0.242||0.604|
The most reliable hosting company in July with no failed requests was New York Internet, which last month ranked third. They opened their first facility in Lower Manhattan in 1996, and also run a datacentre 40 miles away in New Jersey for business continuity and backup operations. Within our million busiest websites, Opera's FastMail service is the most popular website hosted by New York Internet.
Logicworks and Netcetera had the same number of failed requests and are therefore ranked by average connection time in second and third places respectively. Logicworks, which last appeared in the top ten in February in seventh position, offer cloud computing and managed hosting to some prominent brands including Dow Jones and NBC. Netcetera appeared in first place last month, however despite slipping slightly to third they actually experienced a drop in the number of failed requests this month.
Swishmail (ranked fourth) and iWeb Technologies (ranked fifth) also had the same number of failed requests, but had connection times of 73ms and 82ms respectively. Both companies have now appeared in the top ten for the fifth time this year.
Linux is the favourite choice of operating system for this month’s most reliable hosts, with six of the top ten running on the OS; three companies used FreeBSD, including first place New York Internet; and one company was running Windows Server 2008.
Netcraft measures and makes available the response times of around forty 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 that the number of failed requests are equal, the sites are ranked by average connection times.
Information on the measurement process and current measurements is available.