You are here: Home > Analyse > Statistics > RIPEstat

Analysis of Japan Earthquake effects - 11 March 2011

On 11th March 2011 at 1446 local time (0546 GMT) an earthquake of magnitude 8.9 was recorded off the coast of Honshu, in the north-east of Japan. A number of other tremors and a tsunami followed.

We looked at data from the routing and connectivity measurements carried out by the RIPE NCC to try to identify the extent of the effect of the earthquake on Internet operations in the country.

Overall, it appears that the impact on the Internet was quite small. The RIPE NCC DNS monitoring service did not see any impact on the .jp top level domain. The RIPE NCC active measurement networks, TTM and RIPE Atlas, did not observe any effects.

Effect on visible prefixes

This graph shows the prefixes visible from our sample set. Based on snapshots taken every 8 hours, around 2% fewer prefixes were visible after the events.

It is worth noting that changes of this magnitude are quite normal on the Internet, and that this observation is by no means indicative of the earthquake impacting Internet operations. This serves to highlight the resilience of the Internet.


Sample of visible prefixes from Japan

Effect on BGP path length

This graph shows the BGP path length visible from our sample set. Based on snapshots taken some hours before the earthquake, and some hours after, the paths became 0.2% longer.

It is worth noting that changes of this magnitude are quite normal on the Internet, and that this observation is by no means indicative of the earthquake impacting Internet operations. This serves to highlight the resilience of the Internet.

Sample of path lengths for prefixes from Japan

Other sources

After the earthquakes, the JPNAP Internet exchange point saw a significant reduction in exchange traffic. It is not known whether this is due to fewer people attempting to use the Internet, or a reduction in the availability of connectivity.

JPNAP showing drop in traffic

Traffic on the JPNAP Internet exchange point between and 10 March 10:15 UTC and 11 March 2011 10:15 UTC

Methodology

The sample of 1514 prefixes we monitor were selected from the 5000 ranges allocated to Japanese organisations, as documented by APNIC here. We searched the RIPE NCC Routing Information Service (RIS) for announced prefixes from within these allocations which were visible between on March 30 2011 and then selected a sample from these prefixes.

The visible prefix graph is constructed from snapshots of routing information, taken every 8 hours.

All routing data is taken from the RIPE NCC's RIS, which collects global Border Gateway Protocol (BGP) information observed from over 600 networks worldwide.