24th, Dec.2004
33,979 km connection from Tokyo through Chicago, Amsterdam, NewYork to Tokyo Latency 499ms RTT
Network Configuration
treceroute from Opteron1 to Opetron3 (Tokyo - IEEAF - CAnet*4 - Surfnet - Abilene -JGN2)
[root@opteron1 root]# traceroute -q 1 opteron3c
traceroute to opteron3c (192.31.124.5), 30 hops max, 46 byte packets
1 vlan911.sl-e1200.sl.startap.net (145.146.101.1) 144.016 ms
2 vlan10.AR5.Amsterdam1.surf.net (145.146.100.193) 243.361 ms
3 PO6-0.CR1.Amsterdam2.surf.net (145.145.164.1) 260.205 ms
4 PO15-0.CR1.Amsterdam1.surf.net (145.145.160.17) 243.491 ms
5 nycmng-oc192-surfnet.abilene.ucaid.edu (198.32.8.104) 317.529 ms
6 chinng-nycmng.abilene.ucaid.edu (198.32.8.82) 337.563 ms
7 transpac-chinng.jp.apan.net (192.203.116.1) 357.193 ms
8 tpr4-transpac.jp.apan.net (203.181.248.218) 576.975 ms
9 opteron3c (192.31.124.5) 498.959 ms
treceroute from Opteron3 to Opetron1 (Tokyo - JGN2 - Alibene - Surfnet - CAnet*4 - IEEAF)
[root@opteron3 root]# traceroute -q 1 opteron1c
traceroute to opteron1c (145.146.101.4), 30 hops max, 46 byte packets
1 vlan914.tpr4.jp.apan.net (192.31.124.1) 68.873 ms
2 transpac-tpr4.jp.apan.net (203.181.248.217) 182.418 ms
3 chinng-transpac.abilene.ucaid.edu (192.203.116.2) 161.681 ms
4 nycmng-chinng.abilene.ucaid.edu (198.32.8.83) 181.687 ms
5 po5-0.br1.Amsterdam1.surf.net (198.32.8.105) 256.001 ms
6 PO0-0.AR5.Amsterdam1.surf.net (145.145.162.2) 256.115 ms
7 vlan10.sl-e1200.sl.startap.net (145.146.100.194) 358.904 ms
8 opteron1c (145.146.101.4) 498.940 ms
Internet2 TCP Single stream Land Speed Record measurement
System
- Server configuration
- CPU: Dual AMD Opteron 248, 2.2GHz
- Mother Board: RioWorks HDAMA rev. E
- Memory: 1G bytes, Corsair Twinx CMX512RE-3200LLPT
x 2, PC3200, CL=2
- OS: Linux kernel 2.6.6
- Disk Seagate IDE 80GB, 7200r.p.m. (disk speed not
essential for performance)
- Network interface card
- Chelsio T110 (10GBASE-SR), TCP offload engine (TOE),
- PCI-X/133 MHz I/O
bus connection
- TOE function ON
- Technical points
- Traditional tuning and optimization methods congestion window size,
buffer size, size of queue etc. Ethernet frame size (standard frame, jumbo
frame)
- Tuning
and optimization by Inter-layer coordination Fine-grained pacing by offload
engine optimization at slow start phase Utilization of flow control
- Target Performance of single stream TCP on 30,000 km circuit
- 7.21 Gbps (TCP
payload), standard frame -> 216,300Terabit meter / second (Internet2 LSR approved Single/Multi TCP Stream record)
Results
Traffic Graphs
Name of Institutions and persons
- The University of Tokyo
- Kei Hiraki
- Mary Inaba
- Makoto Nakamura
- Junji Tamatsukuri
- Nao Aoshima
- Ryo Nishimura
- WIDE project
- Akira Kato
- Seiichi Yamamoto
- Chelsio communications
- Felix Marti
- Wael Noureddine
- Fujitsu Computer Technologies LTD
- Ryutaro Kurusu
- Masakazu Sakamoto
- Yuki Furukawa
- Yukichi Ikuta
- NTT Communications
- Katsuyuki Hasebe
- Miyuki Neo
- Yu Inada
- Mitsuo Murakami
- APAN
- Kazunori Konishi
- Yoshitaka Hattori
- Jin Tanaka
- Satoshi Matsui
- JGN2
- IEEAF
- Tyco Telecommunications
- Northwest Giga Pop
- CANARIE
- Rene Hatem
- Damir Pobric
- Thomas Tam
- Herve Guy
- Carleton University
- StarLight
- SURFnet
- Kees Neggers
- Erik-Jan Bos
- Dennis Paus
- Roeland Nuijts
- SARA
- University of Amsterdam
- Cees de Laat
- Paola Grosso
- Freek Dijkstra
Supported by
- Special Coordination Fund for Promoting Science and Technology, Ministr
y of Education, Culture, Sports, Science and Technology, Japan
- Foundry Networks, Inc.
- CISCO systems, Inc.
- NetOne Systems Co., Ltd.
- ClearSight Networks
- TOYO Corporation
