Blogger maintenance causes extended outage

Blogger is currently experiencing an extended outage after Wednesday's scheduled maintenance introduced some problems. The site was taken down again at approximately 20:00 UTC on Thursday and is still unavailable to blog authors.

Blogger is running in read-only mode while the problem is resolved. This means blogs can still be visited, but authors cannot write new posts and visitors cannot submit comments.

The Blogger Status blog, which is itself hosted on the Blogger platform, has not been updated since Monday. The most recent post read, "Blogger will go into read-only mode Wednesday (5/11) 10:00PM PST for about an hour for maintenance".

There has been speculation that the current outage was caused by a new user interface being rolled out; however, Blogger confirmed through its Twitter stream that this was not the cause.

Some users have expressed anger at losing posts and comments. A Google employee confirmed that Blogger had rolled back to a previous maintenance release, causing all posts and comments made after 7:37 am PDT on May 11 to be removed. The most recent tweet from Blogger suggests that these have only been temporarily removed.

Most Reliable Hosting Company Sites in April 2011

Rank Company site OS Outage
hh:mm:ss
Failed
Req%
DNS Connect First
byte
Total
1 Rackspace F5 Big-IP 0:00:00 0.008 0.120 0.063 0.127 0.127
2 www.qubenet.net Linux 0:00:00 0.015 0.081 0.043 0.088 0.088
3 www.netcetera.co.uk Windows 0:00:00 0.015 0.067 0.071 0.143 0.288
4 Datapipe FreeBSD 0:00:00 0.019 0.124 0.008 0.019 0.026
5 New York Internet FreeBSD 0:00:00 0.019 0.167 0.063 0.126 0.341
6 www.logicworks.net Linux 0:00:00 0.027 0.114 0.064 0.142 0.344
7 INetU unknown 0:00:00 0.031 0.097 0.039 0.101 0.242
8 Hosting 4 Less Linux 0:00:00 0.035 0.122 0.096 0.195 0.422
9 www.serverbeach.com Linux 0:00:00 0.039 0.076 0.007 0.047 0.080
10 www.poundhost.com Linux 0:00:00 0.039 0.215 0.061 0.136 0.260

See full table

Heading the table for April with only two failed requests from any of the performance monitors during the month was Rackspace. Rackspace provides managed and cloud hosting from nine datacentres in the U.S., the U.K. and China. Rackspace frequently features in Netcraft's top ten most reliable hosting companies, having appeared five times in the last year.

Second most reliable this month was Qube, a London-based hosting company which also has datacentres in New York and Zurich. Qube provide managed hosting, managed colocation and cloud hosting for a wide variety of customers, particularly in the areas of finance and new media. Although Qube has previously appeared in the top ten, this is the first time the hosting company has made it to the top three.

The third most reliable hosting company in April was Netcetera. Netcetera experienced the same number of failed requests as Qube, but had a longer average connection time. Netcetera provides a wide range of colocation, hosting and cloud services to customers throughout the world.

Five of the top ten sites this month were running on Linux, two were running on FreeBSD, and one on Windows. Rackspace's site, which performed best this month, is hosted on F5 Big-IP.

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 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.

Fire at data centre sends Aruba offline

Italian hosting company Aruba was knocked offline for a few hours today after a fire broke out in the centre of a server farm.

Aruba said the fire involved UPS batteries and confirmed that no servers had been damaged; however, the fire alarm system caused the power to be cut, sending many websites offline. Aruba started restoring power once it was deemed safe to do so.

At the time of writing, Aruba has restored 2 out of 3 data rooms and warns that the ongoing UPS restoration may result in unexpected downtime if there are any further power interruptions. Further updates can be found on Aruba's Twitter page.

Most Reliable Hosting Company Sites in March 2011

Rank Company site OS Outage
hh:mm:ss
Failed
Req%
DNS Connect First
byte
Total
1 www.dinahosting.com Linux 0:00:00 0.011 0.241 0.077 0.154 0.154
2 Virtual Internet Linux 0:00:00 0.030 0.136 0.095 0.192 0.393
3 iWeb Technologies Linux 0:00:00 0.034 0.077 0.047 0.094 0.094
4 Kattare Internet Services Linux 0:00:00 0.034 0.095 0.073 0.146 0.296
5 Datapipe FreeBSD 0:00:00 0.037 0.105 0.023 0.047 0.066
6 Rackspace F5 BIG-IP 0:00:00 0.037 0.212 0.234 0.282 0.282
7 INetU unknown 0:00:00 0.041 0.065 0.030 0.375 0.491
8 www.logicworks.net Linux 0:00:00 0.041 0.145 0.048 0.515 0.671
9 www.qubenet.net Linux 0:00:00 0.041 0.114 0.056 0.113 0.113
10 Swishmail FreeBSD 0:00:00 0.045 0.135 0.049 0.097 0.261

See full table

The most reliable hosting company in March was dinahosting, which has been offering hosting services since 2002. The company's data centre is spread across 25 buildings located in Madrid, Spain, which are physically protected by sluice gates and biometric access controls. Besides Spanish, dinahosting also provides support for customers who speak Catalan, Galician, English, and Portuguese. To increase performance from other countries, dinahosting also has additional DNS servers in London and Dallas.

dinahosting is currently offering 50% off its RealCloud cloud hosting service, which is managed by Xen hypervisor 4.0.0 running on Dell PowerEdge R410 servers. This platform supports auto-scaling to dynamically allocate additional resources when required, such as during traffic spikes. As with many other cloud hosting services, CPU-hours and bandwidth are charged for on a pay-as-you-go basis.

Virtual Internet took second place, with only eight failed requests throughout March. The UK-based company is scheduled to launch its flexible managed hosting and cloud hosting solutions in the USA from 1st May 2011. Virtual Intenet's cloud hosting services are based on VMWare, and include a 100% uptime guarantee, automatic crash recovery, 1 gigabit networking and 20 day try-before-you-buy offer. Other services offered by Virtual Internet include colocation, global content delivery and business email hosting.

Six of the most reliable hosting company sites in March were running on Linux, including each company within the top four. Of the remaining companies, two used FreeBSD and one used F5 BIG-IP.

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 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.

LiveJournal under DDoS attack

LiveJournal has been knocked offline by another DDoS (distributed denial of service) attack, less than a week after a separate sustained attack caused the site to go down for several hours. In response to last week's attack, LiveJournal upgraded their servers to make the site run faster; however, this does not appear to have prevented the current attack from succeeding.

Svetlana Ivannikova, Head of LiveJournal Russia, confirmed that the current outage was caused by another DDoS attack: "We can confirm that the service has not been working correctly for the last hour due to another DDoS attack on LiveJournal. Administrators are aware of the problem and trying to identify the source and target of the attack". No further details were given at this time.

The attack which caused last week's outage apparently began on 24th March, but LiveJournal largely withstood the attack until it was ramped up on 30th March. LiveJournal maintenance said, "Turns out we upset our attackers and they started hitting us 10x harder".

Both www.livejournal.com and news.livejournal.com were still inaccessible at the time of publication.

Faulty switch sends BBC offline

The whole of the BBC's public-facing network disappeared from the internet late last night. Although the outage only lasted for around an hour, the unavailability of popular sites such as BBC News and iPlayer caused an eruption of comments and complaints on Twitter and other social networking sites.

Tony Ageh, the Controller of BBC Archives, broke more than 4 months of silence on Twitter to express his frustration. Fellow employee Mo McRoberts responded, saying that "somebody responsible for our routers did something very silly indeed. took out the whole lot." He later confirmed that the outage was not caused by a denial of service (DoS) attack. (McRoberts' tweets are his own, and not necessarily those of his employer).

The BBC's technology correspondent, Rory Cellan-Jones, summarised the widespread impact: "Love the terse bulletin on last night's BBC web failure. Cause of issue: faulty switch. Services impacted: Everything.."

The BBC News website suffered similar outages back in 2007 when a routine software deployment caused some unforeseen performance issues.