The M4.1 and M4.8 events on December 2, 2000 provide an interesting contrast in web server traffic in response to felt earthquakes.
December 2, 2000 had two felt earthquakes in California. At 00:28:07 there was a M4.1 located near Big Bear City, and at 07:34:15 there was a M4.8 near Truckee. The Big Bear event is in the Southern California area served by the USGS Pasadena office, and it is normal for felt events in Southern California to generate web traffic surges on our server. The Truckee event is in Northern California, so the traffic it generated on the Pasadena server was due to people coming in to fill out the Community Internet Intensity Map [CIIM] questionnaires. These two mechanisms for bringing visitors to our site generate distinctly different patterns of traffic.
|Web Server Traffic: 2 Dec, 2000|
Here is the overall traffic pattern for the 24 hours, in 10-second chunks. Note the two large peaks. The first is the M4.1 Big Bear event, which was felt in parts of Southern California. The second event was the M4.8 near Truckee. This was not felt in Southern California, but links from the USGS Menlo Park web page brought people to the Pasadena server to fill out questionnaires about the event.
Normally, a felt event in the middle of the night will generate an immediate traffic spike which decays rapidly. Then a second less-peaked swell comes after dawn, when people who felt the earthquake wake up and go looking for information about it. In this case, the "dawn swell" has a second peak added on top of it.
Here is detail for the 20 minutes surrounding the first peak. This is the surge that came from people who felt the M4.1 Big Bear event. Note that there is about a one minute lag from the event origin time to when there is an unmistakable increase in traffic. This is probably about the amount of time it takes a person who is already at their computer to find a bookmark for our page. The "Recent Earthquakes" pages are generally updated within about 2-4 minutes after an event, so some people are coming to the web site before there is even anything to see.
This is detail of the 20 minutes around the beginning of the second surge. This was people coming to fill out CIIM questionnaires about the M4.8 Truckee event. Note that the increase is not as steep as in the first peak. This is most likely because people coming to the Pasadena site were referred from the Menlo Park site. This made us the second stop instead of the first, and different people will take different amounts of time to read the first page before following the link.
|Recent Earthquakes: December 2|
This is day's traffic for just the "Recent Earthquakes" pages. Note that this generated a lot of activity right after the first event, but then there was no activity until dawn. This indicates that most people who came to the site right after the event had their curiosity satisfied, and the rest of the people who felt it didn't feel the need to check it until they woke up in the morning.
|CIIM: December 2|
|Recent Earthquakes: 00:20-00:40|
As above, this is detail of the "Recent Earthquakes" traffic for the 20 minutes around the first peak. There is a lag of not quite 90 seconds before the first visitor went to the "Recent Earthquakes" page. The people who were first on the site would not have found the event there yet.
This is detail of CIIM traffic around the first event. Note that the lag before the first vistor is nearly two minutes here. This is probably because this is again a secondary destination for people looking for information about an event they just felt.
|Recent Earthquakes: 07:25-07:45|
And this is the 20-minute detail for the "Recent Earthquakes" pages around the second surge. Note that this event didn't really generate a peak at all. This is most likely because the people coming to the Pasadena server after this event were coming to fill out a CIIM questionnaire. It is most likely that they were referred here from the "Recent Earthquakes" pages on the Menlo Park server, so they had already seen the map.
|CIIM Questionnaires: December 2|
This is the graph of CIIM questionnaires submitted for the day. The important features in this graph are just to see that the two lines are identical. This indicates that the web server was able to keep up with the load. In the past, we have had situations where the server was not able to process the questionnaires as fast as they came in. In this case, we had capacity to spare. The peak rate was 16/min, and our server is capable of processing 1200/min. So in this case, we were able to deal with the load comfortably.