**3.2 Before the earthquake**

364 Earthquake Research and Analysis – Statistical Studies, Observations and Planning

This complicated topology made the operation of the R&E networks more difficult. To simplify operations, Research and Education Network Operators Group (RENOG), (Research and Education Network Operators Group [RENOG], n.d.) was developed and started reconfiguring the complicated and unstable routing. However, in the APAN area, the complicated routing unexpectedly worked and it was able to maintain high-speed communication with most of the network researchers, though with some delay. Network engineers like the word "redundancy" but most of the network engineers in the APAN area did not expect that the complicated topology and the complicated routing information

Fig. 1. Star shaped R&E networks(Robb, 2006)

Fig. 2. Actual R&E networks(Robb, 2006)

would work so well in an emergency.

The TEIN2 project started in 2006. Before that, the network topology style in Asia Pacific area was close to the star shape. Most of the networks in the APAN area started in Japan, and if a network did not start from Japan, it started from one of the point sites. That is why the star shape was kept. However, the TEIN2 topology does not look like a star. Even inside TEIN2, there are two or three routes to reach any other sites. An engineering meeting was held before TEIN2 started at which an agreement on routing policy seemed to have been reached. However, after TEIN2 started, there happened some of routing troubles in Asian Pacific area. A few troubles were from TEIN2 network directly but others were the routing advertise issues from the National Research and Education Networks (NRENs) of the TEIN2. To solve these problems, the operators held many meetings to develop tools. One tool summarized traffic conditions for each route (Asia-Pacific Advanced Network Japan [APAN-JP], n.d.). Another was a database to record preferable routes for each XP (Kurokawa, 2006). These were essentially monitoring and advising tools, not route configuration tools. Actual operations were done at the XPs based on the database, which was built by the database tool (Kurokawa, 2006). At the APAN Tokyo XP, the abstract of the routing policy was as follows.


Figure 3 shows the one of the worst examples of communication between US university through KR. That is, if a university in US established communication with Merit or the University of Michigan, the packets were transferred through KR. In such a case, TCP-based applications often met with communication problems.

Fig. 3. Incorrect routing between universities in US through KR(Robb, 2006)

Experience with Restoration of Asia Pacific Network Failures from Taiwan Earthquake 367

the group that consisted of JP, KR, TW, and US, and the other consisted of CN, HK, VN, MY,

TH, SG, ID, and PH (Fig 5).

Fig. 5. Splitted R&E networks in Asia Pacific area

2. Lost primary BGP peerings

Fig. 6. Traffic weather map on Dec.27 2006

The Internet disconnection occurred in the following order. 1. Link-layer disconnection because of the fiber cut

3. Automatic BGP re-routing along the alternative peer if any
