Today we're looking at some server overload circumstances from the sad, to the silly and the very expensive ... and some where the hackers got the blame for something that may well have been the load tester's responsibility!

Website flooded after heavy rains

The UK Environment Agency set up a National Floodline in 2002, to help provide residents information about levels of flooding in their area and whether they should stay, or evacuate. This potentially life-changing information was unavailable for three days - naturally, over a period of extremely heavy rainfall. Both the servers and people's homes were being flooded! The website developers hadn't anticipated that there would be a natural very low, then very high interest in the site, and had load tested for an average use scenario. Luckily, the lack of information didn't cost anybody their lives.

Expert PR turns lack of load testing into a positive

Early in 2009 when Microsoft released the beta version of Windows 7 (a free download), there was massive interest. So massive, that it was only available for a short while on the day of release before the servers overloaded. In Microsoft style, though, there was no apology. "The fact we're seeing such enthusiasm, that's just wow", said Mike Nash, corporate vice president. The website itself said "Stay tuned! We are excited that you are excited!".

$16,000 per minute web glitch

They may not have been sure what the problem was - but it cost Amazon $16,000 a minute, based on their global revenue figures for the year before. A 'non-security related problem' caused a 1.1 Service Unavailable message at Amazon in June 2008. It was only the American site that was unavailable - China's, France's, the UK's and Germany's sites were still up. And the damage still ran into the hundreds of thousands...

Google Docs not so reliable after all

Google has been pitching their Docs function as a credible, reliable alternative to using Microsoft Office and other PC-based software. Cloud computing may be the way of the future, but Google proved that it still isn't a reliable way for the present, when Docs stranded user files in the cloud for around an hour. Luckily the document backup system is still one of the most rigorous on the web, so no actual data was lost. The damage to their reputation was significant, though.

Web app overload a sorrowful circumstance

Bill Shunn is a freelance programmer who created a survivor list for the Web in the days after the 9/11 bombings in New York. The massive response, with no advertising and no hype, still managed to crash his servers after just over 24 hours in business. A sad situation made all the worse by a lack of traffic capacity and resources for his and other websites - but the solo operators trying to make a difference are commended anyway.

Overload or DDOS?

Hasbro made much ado about the Facebook app version of their game, Scrabulous, suing it off the web in mid-2008. However, when the official Hasbro version arrived (in beta), it crashed immediately afterwards. It is unclear whether it was hacked off the web, as the creators claim, or whether Scrabulous users just couldn't wait to get their letter-jumbling fix and overloaded the servers.

Sources:

Disaster Web Sites Overloaded with Queries About Loved Ones

EA: Hack took Facebook 'Scrabble' down

Amazon suffers U.S. outage on Friday

Google Docs goes down, user data does not

10 Worst Web glitches of 2008 (so far)