December 18, 2012
Hurricane Sandy Was Responsible For Mass Internet Outages
Peter Suciu for redOrbit.com — Your Universe Online
October´s Hurricane Sandy is at present the second-costliest Atlantic storm behind 2005´s Hurricane Katrina. The storm, which developed as a tropical wave in the western Caribbean Sea on October 22 moved ashore near Atlantic City, New Jersey on October 27.
These are the findings of a new report released by the Information Sciences Institute (ISI) at the USC Viterbi School of Engineering, and funded by the U.S. Department of Homeland Security Science and Technology Directorate.
The findings were made available in a report released on Tuesday, while the raw data from the study will be made available to other scientists who would like to analyze it further.
And the information released so far shows that the storm was in fact responsible for that spike.
USC researchers who track Internet outages across the world have found that twice as much of the Internet was down in the United States as usual. While it is common for 0.3 percent of the Internet to be down at any point on any given day, just prior to the storm that number was actually lower at 0.2 percent. This number has even been called good by global standards.
However, once the storm made landfall that number jumped to 0.43 percent and it took about four days for it to return to levels that would again be considered normal.
“On a national scale, the amount of outage is small, showing how robust the Internet is. However, this significant increase in outages shows the large impact Sandy had on our national infrastructure,” said John Heidemann, research professor of computer science and project leader in the Computer Networks Division of ISI.
Heidemann led the team that tracked an analyzed the data, and has worked with graduate student Lin Quan and research staff member Yuri Pradkin, both also from ISI, in an experiment that helped track the outages.
The team sent tiny packets of data — or “pings” — to networks and waited for the responses — “echoes.”
While it is common that some networks will not respond, notably those with a firewall in place, the method is typically used to determine if there is a connection to a specific machine or network. This method has thus shown to provide a statistically reasonable picture of outages. If the ping gets a response then the network sector is active; no response may suggest that there is an outage.
In the case of Sandy the ISI team was able to pinpoint where the outages were occurring, showing a spike in outages in New Jersey and New York after Sandy made landfall. The results of this research were published as a technical report on the ISI website.
The data doesn´t yet show how many individuals were actually affected by the outage, but it does shed some light about the scale and location of the outages. While it may seem obvious that the storm caused the outage, this data would help Internet service providers better prepare for the next storm.
It could also help ISPs determine how to best allocate resources in response to a natural disaster.
“Our work measures the virtual world to peer into the physical,” Heidemann said in a statement. “We are working to improve the coverage of our techniques to provide a nearly real-time view of outages across the entire Internet. We hope that our approach can help first responders quickly understand the scope of evolving natural disasters.”