**3. Background of R&E network**

#### **3.1 R&E network status in 2006**

In the old star-shaped R&E network topology (Fig. 1) communication between the point sites had to go through the center of the star. Communication along these routes was frequently delayed due to the configuration. Now, however, the R&E network community has grown, and the former point sites are now sometimes the center of a star. In some cases this growth generated several routes between two sites (Fig. 2).

Experience with Restoration of Asia Pacific Network Failures from Taiwan Earthquake 365

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

Communication with ID, SG, MY, TH, VN, CN, and AU (Tbl. 3) should be routed

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

Communication with Micronesia should be routed through the Hawaii XP.

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

**3.2 Before the earthquake** 

routing policy was as follows.

Communication with PH should be direct.

Communication with TW should be direct.

Communication with US should be direct.

Generic communication with EU should be through US.

applications often met with communication problems.

through TEIN2.

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

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

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 would work so well in an emergency.
