1. KEK-CERN JEG experiment CHEP2000 conference at Padova on 9 February 2000 Yukio.Karita@kek.jp http://www.kek.jp/~karita/chep2000-jeg.ppt

2. Co-authors Institutes and people who participated in this experiment are: F KEK: Yukio Karita, Soh Suzuki, Fukuko Yuasa, Youhei Morita, Takashi Sasaki, Katsuya Amako, Teiji Nakamura, Kiyoharu Hashimoto, Hiroyuki Sato, Yoshiyuki Watase F CERN: Olivier Martin, Paolo Moroni, Jiri Navratil F CRL (Communications Research Laboratory): Fumito Kubuta, Hideki Otsuki, Hiroaki Harai, Shin’ichi Nakagawa, Akihiko Machizawa F TAO (Telecommunications Advancement Organization of Japan): Hideo Miyahara, Isao Kaneko

3. What is JEG? F JEG (Japan-Europe Global Architecture for Multimedia Access) Project –Joint project of MPT (Ministry of Post and Telecommunications in Japan) and ESA (European Space Agency) –The objective of the project is to create a satellite-based infrastructure that will allow the European and Japanese users to exploit the connectivity created for the development and trial of new Applications.

4. INTELSAT Earth Station used at JEG Leuk

5. Some history of JEG F JEG started and a 2Mbps Japan-Europe circuit became available mid 1997. F 1997-1998, JEG was used as TDM base. F The TDM multiplexors used were located in the carriers, and thus the configuration change was not so flexible and took time. F When the KEK-CERN joined the JEG, we studied what scheme we should have.

6. JEG outline topology as of 1997-1998

7. JEG Scheme since 1999 F We changed the JEG topology as follows: –Totally ATM based. –All the projects are to be connected to CRL in Japan and to CERN in Europe by ATM. –Configuration of the ATM switches in CRL and CERN are easily changed. CRL CERN KEK CERN NCC IGR KEIO INRIA 2Mbps

8. Outline Topology of JEG Network since 1999 CRL Leuk E/S KEK NCC IGR: Institute of Gustav Roissy at Paris NCC:National Cancer Center at Tokyo CERN F Tele-medicine applications: NCC/CRL ⇔ IGR F High-energy physics applications: KEK ⇔ CERN CERN IGR KEK NCC INTELSAT 2 Mbps Yamaguchi E/S

9. F Planned tests on the KEK-CERN JEG –monarc –h323 –ipv6 Three ATM PVCs are configured to do the planned three tests independently. However “ipv6” was not tried at the phase I. And “monarc” and “h323” were tried on one environment. So only one PVC was actually used.

10. Cisco ATM switch ATM switch JEG F KEK’s LAN and CERN’s LAN are connected with JEG’s 2Mbps ATM PVC. 2Mbps LAN KEKCERN Routers are configured so that only some specific traffic is routed over the JEG circuit.

11. F Some characteristic of the JEG circuit –large but constant RTT: 660ms –guaranteed bandwidth: 2Mbps F This gave very good environment for “monarc” and “h323”.

12. Router configuration at KEK F enabled each host to choose at anytime which route to take between the JEG testbed and the production route, by a command in the host: route add -host x router[1,2], where x is a host in CERN. F we wanted to have the same environment in CERN.

13. h323 PictureTel’s LiveLAN -very good at 192Kbps -some problem at 384Kbps or more -some improvements are being discussed -used as a base for the collaboration in “monarc”

14. monarc F tried to demonstrate the capability of world-wide distributed database (replication, WAN access, paralleled I/O etc) F tried to validate the MONARC simulation program F tried to understand the performance of the database over the wide area network F tried to identify the bottlenecks and possible improvements of the distributed database

15. monarc (contd.) F Performance of Data Replication and AMS in Objectivity/DB via WAN (JEG) was measured, and was compared with the measurement in LAN. F Found that Data Replication works well in LAN but not efficiently in WAN, especially NOT at large RTT. F Some improvements in AMS was proposed to Objectivity Inc. F Details are reported by H.Sato at D235 “Evaluation of Objectivity/AMS on the Wide Area Network”

16. Present and Future Plan July 97Mar 98Sep Nov98July 98Aug 992003? JEG Project Phase 2 Phase 3 IDC98 Earth Observation Internet Measurements Launch of WINS & GCTS Videoconference Tele-Medicine Global 360 (IST98) High-energy physics ?

17. Satellite Internet Applications (Proposal for 2nd Phase Project) F 1. Joint Project Framework –Japan: MPT/Sat-Com Application Experiments Promotion Conference –Europe : ESA/5th Framework Project? F 2. Objectives -Next-generation Satellite Internet Applications - : Huge data distribution/co-owner -Broadband multicast satellite down link -Narrow terrestrial return link -Inter-continental backbone -Distributed Databases: - Earth environment data - High-energy physics data Inter-continental backbone Regional Satellite Access Networks Access Networks

18. Summary –Separately from the production networks, a testbed named JEG (Japan-Europe Gamma) was provided to us by MPT and ESA during June-November 1999. –ATM PVCs connecting KEK and CERN (2Mbps, satellite) for testing MONARC and H323 have worked and produced many data without being affected by the ordinary traffic. –The first phase was terminated on 30 Nov 1999, then the second phase will start early this year with some more bandwidth.