The M7.1 Hector Mine earthquake was felt over all of Southern California, and generated a huge surge of traffic to the USGS Pasadena web server.
|Web Server Traffic: 16 Oct, 1999|
The Hector Mine earthquake on October 16, 1999 was the first big event to completely overwhelm our web server. The story of its effect on the server and what we did in response is told in "Web Servers, Earthquakes, and the Slashdot Effect".
Here is the overall traffic pattern for the 24 hours. The hole from 09:00-11:45 was caused by the log disk on the web server filling up. Traffic was being processed, but not logged during this time.
Because this event occurred in the middle of the night, not everyone jumped out of bed and ran to their computers. Those who did made up the first big spike, and the rest of Los Angeles' web surfers made up the second swell after dawn. This event was the first time that we observed this effect, but it has happened after several other middle-of-the-night events.
Here is detail for the first 60 minutes after the event. Note how the traffic rate sagged from 03:10 to 03:15. Also note the purple line, which represents completed Report an Earthquake questionnaires. Each questionnaire has to be processed by a CGI script on the server. The USGS office server at that time was capable of processing about one questionnaire every three seconds. The input rate here was about 1 questionnaire/second, and the server was brought to its knees. This kept it from servicing regular HTTP traffic.
This is detail of the 20 minutes around the beginning of the big surge. Note that the lag from between the event origin time and the beginning of the traffic surge is almost two minutes. This is probably due to the fact that the S waves took about one minute to reach Los Angeles, and then one more minute for web surfers to find their bookmarks.