Flickr outage reveals site's scale
Doubtless many Flickr customers were not happy with the site's problems Monday, in which the Yahoo-owned site displayed the wrong images. Perhaps more interesting than the explanation of what went wrong, however, was the revelation of the volume of traffic the photo-hosting site handles.
That's a lot of photos--more than 11,574 per second on busy days.
Costello also took pains to apologize for the problems and assure Flickr users that no files were corrupted and that the site wasn't hacked. Instead, the problem lay with flawed information served up by caching servers, an army of helper machines that accelerate distribution of data housed on primary servers.
"Tonight's problem was a result a few of the photocaches going berserk and instead of returning the correct image file when a particular photo was being requested, it just returning some random image that happened to be in the cache," Costello said. "To be clear, we regard this as a serious problem, but it is something that goes away as soon as we restart the malfunctioning servers (Tonight we found that the servers were going insane again shortly after restarting, but we have isolated the problem and believe we have a permanent fix)," Costello said.