The Great Michael Jackson Web Collapse Downgraded to “Stumble”
by Peter Kafka
Posted on June 29, 2009 at 10:12 AM PT
We’ve previously noted that the Web is great at transmitting information quickly, though not always accurately. Same goes, apparently, for stories about the Web’s ability to transmit information quickly.
After Michael Jackson died on Thursday, we saw a rash of stories about the Internet’s inability to handle the crush of traffic the news event generated. Many of those stories cited the same source to illustrate the problem: A report Web performance consultants Keynote Systems (KEYN) issued that day that said Time Warner’s (TWX) AOL, Disney’s (DIS) ABC and CBS (CBS), among others, experienced “marked slowdowns in performance.”
I found that a little bit surprising since my desultory and highly unscientific survey of the Web found that most sites were working pretty well as the news broke Thursday afternoon. And to its credit, Keynote’s report never said the sites failed. Instead, it said average downloading times at news sites doubled, to nearly nine seconds, and that “average availability of sites” went from 100 percent to 86 percent.
But even those nondeaths were greatly exaggerated, Keynote now says. In a report issued late Friday, the service says that its analytics were measuring the wrong thing on Thursday. The news sites themselves performed OK, but the third-party networks serving ads frequently failed to keep up. Oh, and the two-hour failure it reported at ABCNews.com never happened.
To be sure, some individual Web sites and services did trip a bit when confronted with a rush of Jackson traffic. But I’ve yet to hear of a service that completely went down. Google News (GOOG) gave some users an error message for 45 minutes, but the main search service still seemed to function. AOL’s AIM went down, but not the rest of the site. Twitter got slower but still stayed up. Etc.
Even more comforting: I have yet to hear of the Web’s infrastructure itself buckling under traffic. Anyone have any evidence to the contrary? Let me know.
[Image credit: University of Washington Librarires via pri.studio360]
