BBC News Knocked Offline by Performance Changes

The BBC News website has been suffering outages today while its hosting location fluctuated between BBC Internet Services and the Akamai web application acceleration and performance management service.


For many users this afternoon, the front page of the BBC News site has been slow to respond, often displaying error messages such as "No suitable nodes are available to serve your request" and "an error occurred while processing this directive". Other users found that requests to caused the server to continually redirect their request back to, causing an infinite redirection loop which would have added to the load on the servers.


Today's performance problems coincide with apparent moves to and from the Akamai content distribution network. Prior to today, the site had been self-hosted by BBC Internet Services in Docklands, London.


For some periods today, the BBC News website had resolved to IP addresses belonging to Akamai, while other times it had either been pointed back to BBC Internet Services at Docklands, or did not resolve at all, thus leaving the site completely inaccessible.

Akamai transparently mirrors content stored on web servers and users then access the content from these instead of the origin server. By automatically picking a mirror server that is near to the user, performance is generally increased while decreasing the load on the origin server.

Update: Steve Herrmann, editor of the BBC News website, has published a blog article about the site problems.

INetU, New York Internet and Acens are the Most Reliable Hosting Companies In August 2007

Ranking by Failed Requests and Connection time,
August 1st - 31st 2007


INetU Inc., New York Internet and Acens are the most reliable hosting company sites for August 2007, followed closely by Easynet,iPowerWeb and

This month's leaders include two managed hosting providers who have made frequent appearances in our ranking of the most reliable performers. New York Internet, a colocation and managed services provider in Manhattan, topped the charts for the third time this year. NYI previously was most reliable in January and May of 2007. INetU, a managed hosting provider in Allentown, Pa., was second in the July survey and has appeared in the top 10 many times, including a run of seven straight months in 2004. This is the first appearance on the reliable hoster list for Spanish hosting company Acens. In January private equity firm Nazca Capital bought a major stake in Acens, which had revenue of $18.8 million in 2006.

Close behind the top three are American discount specialists iPowerWeb and Go Daddy, who offer shared hosting accounts starting at $4 to $6 per month. Go Daddy is also the world's largest domain registrar.

Three of the ten most reliable hosts run their web sites on FreeBSD, three on Windows and two on Linux.

Continue reading Hit By Downtime was offline for an hour early Wednesday as Google's popular service encountered performance problems. Blogger is one of the oldest and largest blogging services, hosting several million free blogs. Google "apologized profusely" for the service interruption.

"Blogger and Blog*Spot had an unexpected outage for an hour this morning, starting around 7AM PDT," Google reported on its Blogger status page. "As of 8AM, all but a few blogs are working properly again." The company didn't offer any additional details on the cause of the outage. The Google search engine and other services were unaffected, as the outage appears to have been isolated to Blogger and

iWeb Technologies is Most Reliable Hosting Company In July 2007

Ranking by Failed Requests and Connection time,
July 1st - 31st 2007


iWeb Technologies Inc. is the most reliable hosting company site for July 2007, followed closely by DataPipe, Swishmail, New York Internet and INetU.

This is the third appearance atop the survey this year for iWeb Technologies Inc, a Montreal-based hosting provider that has grown through acquisition, buying eight other hosting companies since 1998. iWeb offers shared hosting, dedicated servers, colocation and domain registration. It had earnings of $637,000 on revenues of 2.3 million for the quarter that ended March 31.

Datapipe, NewYork Internet and iNetU each make regular appearances on our reliability leaderboard, and continue their strong performance this month. Swishmail, which provides business e-mail services, makes its first appearance in the top 10 this month.

Four of the 10 most most reliable hosts run their web sites on Windows, two on Linux and three on FreeBSD.

Continue reading

Lengthy Downtime Due to Affinity-Hostway Migration

Dedicated hosting customers of Affinity Internet's ValueWeb and brands reported lengthy downtime as a planned outage for a data center migration extended well beyond the scheduled 12 to 15 hour maintenance window. The server move between Affinity's Miami, Fla. data center and a Hostway facility about 270 miles away in Tampa, Fla. began Friday night at 8 pm, but late Sunday night many customers reported that their sites remained offline, with some saying their server outages were nearing 48 hours.

Hostway acquired Affinity Internet in April a deal that combined two hosting services focused on the business market. The U.S. hosting market has been undergoing an active period of consolidation, with WebSite Pros buying and Endurance International merging with IPower in other deals.

The main pages for both Hostway and ValueWeb remained online throughout, as only dedicated hosting customers appear to have been affected. A forum post by a manager indicates that all servers had been connected and powered on in the new location by 2 pm Sunday, but acknowledged that servers remained offline.

Service Outage for Fasthosts

UK hosting provider Fasthosts suffered extended downtime on Monday night, which has been attributed to an electrical problem. Fasthosts, which was bought last year by 1&1 Internet, hosts more than 605,000 web sites, making it one of the UK's top three web hosting operations. Fasthosts told customers that the outage resulted from a problem in the electrical equipment that distributes power to servers. While there was no fire, a fire alarm was triggered, which in some scenarios can prompt staff to cut power to parts of the data center for safety reasons.

Last night's downtime can be seen on this performance chart:


A dynamically updating chart of the web site performance is available. Netcraft offers a web site performance monitoring service that provides similar charts, along with e-mail alerts when an outage occurs.