NCFC Introduction Excerpt from "Current Computing/Networking Status in China" By Cindy Zheng (zhengc@Sdsc.Edu) 7/2/93 In June 1993, I finally made my first trip back to Beijing. Dr. Karin asked me to be his representative, to assess and to collaborate with NCFC (National Computer Networking Facility of China) project which is part of KSDP. I also wanted to find out what people in China are thinking and doing about establishing direct network connections to the internet. All the institutions I visited are very anxious to establish direct connections to internet in the U.S. NCFC is especially ready technically and financially. Multi-protocol routers from DEC will arrive in July. CNC (NCFC Computer Network Center) people have done some investigations on network carriers and providers: JVNCnet, Sprint, South Korea (256k) and Hong Kong (64k) etc. They are continuing on more fact-finding and detail negotiations. There have been a few reports about NCFC project that are widely read and routed on the internet. I will only update the progress made this year. NCFC backbone - fiber optic cables are all in place now, waiting for the network routers to arrive to connect among three campus nets (PUnet, TUnet and CASnet). These routers are multi-protocol routers. Both DECnet and TCP/IP protocols are commonly used in China. Within a local area network, subnet routers are developed by local technical people, using PC. They are just as good as what they could have imported, but cost much less. For example, imported network bridges cost $12,000 each, but locally developed ones cost only $1,500 each, with multi-port, SLIP and multi-media capabilities too. They are also working on routers' enhancements, to carry TCP/IP traffic on ethernet, SLIP and X.25, so remote connection can be established to NCFC via CNPAC which is the domestic communication network provided by PTT (Ministry of Post, Telephone and Telegraph). All three institutions under NCFC (Chinese Academy of Sciences, Peking University and Tsinghua University) have started on the second phase of the NCFC project - application development. CAS-MIS has created 20 databases locally, they also initiated SDB (Science Database) project. There has been total $2.25 million supplemental investment made by the 3 institutions - CAS, PU and TU. CAS expanded their LAN, currently connected 19 of their research institutions and will connect 11 more by the end of this year. Right now the fiber-optic cable for 13 more ethernets has arrived, the engineering design is completed. CAS is in the process of selecting contractors to sign the contracts and start the actual construction. At CAS, 100 mbps FDDI ring is operational and a 10 mbps bridge loop is an on-line automatic back-up. There are about 100 accounts established on CASnet systems and CAS has conducted some user network training. CAS is responsible for CNC (Computer Network Center) and intended to make it a supercomputer center which will be shared by other educational and research institutions. This involves the purchasing of a supercomputer. This main computer system is a major item funded under NCFC project. Because the funding comes from both the World Bank and the State Planning Committee, the bid procedure is double complicated. Starting in 1991 but not until April 1993, was NCFC able to complete the technical bid document and submitted to WB for approval. This is the first major step in the purchasing procedure and it is the hardest one too. The progress has been slow, from what I learned, there are three reasons: 1. the required procedure itself is lengthy, 2. the U.S. restrictions to China on exporting high-end computer systems, 3. distrust among different institutions in NCFC The major obstacle is the U.S. export restrictions to China. In the scientific and research arena of the PRC, there is a definite need for supercomputers. I visited some laboratories within CAS. Many research projects require massive computing power and are now crawling on outdated small systems. The U.S. is mainly concerned with any possible military use of exported supercomputers, and therefore attaches conditions such as the U.S. personnel must stay with the system to monitor its usage. Chinese side has no problem with such conditions, but cannot afford to pay such personnel at the U.S. salary level. The type of systems they can get without such problems are mid-range computers; in the U.S. we categorize them as mainframes. Within NCFC, there are two different opinions on the solution. One is to purchase the best it can get and share it within and beyond NCFC. Another is to purchase many workstations instead and spread them among different institutions. This argument is not only provoked by the difficulty of getting a real supercomputer, but also motivated by the distrust among different institutions. Historically, this is the extension of the argument about whether and where should the CNC be. CAS won a very marginal majority vote (by one vote) on hosting CNC there. Up to now, there are still three possible configurations for the main systems: 1. 1 supercomputer with both vector-scalar functionalities 2. 2 mid-size systems 3. many workstations forming a cluster These arguments and indecision also slowed down the progress in completing the bid documents. I visited Peking University Computer Center (PUCC), talked to Zhang Xinghua, the Director of PUCC, Ren Shoukui, the Director of PU Computer Networks Lab and others. I was invited to talk about internet and SDSC and Ren Shoukui gave a report on PUnet. Peking University Computing Center has a staff of 64. Peking University has one of the top libraries in the country and is leading the effort in developing a library information retrieval system. Its library purchases reach $400,000/year, 80,000 books a year. The software for automated management system is completed and operational on their VAX 11/750. But due to limited computing power of the hardware, there can be at most 8-10 users on-line at a time and access is very slow. Professor Hu Daoyuan is the brain behind TUnet. He showed me the TU Computer Center and explained to me the progresses made on TUnet. TsingHua University's FDDI ring is operational. One third of total 2,000 computers, PCs, are connected now. There is a concentrator connected on the FDDI rings, but no workstations are attached so far because they cannot afford the FDDI interfaces for the workstations. TU has an international email connection via X.400 to Canada's University of British Columbia. The cost is the fee for the leased line plus $0.09/min and $0.008/seg(64B) at 2400 or 4800 baud. This connection is still in its "trial run" period and the current traffic volume is insignificant. At TU, a menu-driven Chinese email system is developed in order to help non-computer, non-English users use email. TU is also leading the effort in multi-media research and development and specifically it is attempting to attach voice with email. Among the three institutions within NCFC, Tsinghua University is most aggressive in commercializing its expertise in computer technology. Currently it has contracted to do domestic users training for MicroSoft and AT&T. It also has turned some of its own research results into computer and network products.