Draft
CCIRN Meeting Minutes
7/3/04
I. Meeting
Attendees
Asia-Pacific
delegation:
Shigeki
Goto (Co-Chair) Waseda Un./APAN
JP goto@goto.info.waseda.ac.jp
Borhan
Mohd Ali MyREN MY borhan@eng.upm.edu.my
Jie An CERNET CN anji@cernet.edu.cn
Xing Li CERNET CN xing@cernet.edu.cn
Simon Lin ASCC TW sclin@sinica.edu.tw
George McLaughlin AARnet AU gmm@aarnet.edu.au
Fay Sheu APAN-TW/NCHC
TW faysheu@nchc.org.tw
Eugene Yeh APAN-TW/NCHC TW cyeh@nchc.org.tw
European delegation:
Kees Neggers (Co-Chair) SURFnet NL kees.neggers@surfnet.nl
John Boland HEANET IE john.boland@heanet.ie
David Foster CERN CH david.foster@cern.ch
Tomaz Kalin DANTE UK tomaz.kalin@dante.org.uk
Peter Kirstein UCL UK kirstein@cs.ucl.ac.uk
Karel Vietsch (Info. Coord.) TERENA EU vietsch@terena.nl
David West DANTE
North-American
delegation:
George Strawn (Co-Chair) NSF
Grant
Miller (Info.Coord.) NCO US miller@ccic.gov
Heather Boyles Internet2 US heather@internet2.edu
Jaqueline Brown Pacific Wave US jbrown@cac.washington.edu
Doug Gatchell NSF
David Lassner Hawaii
Un. US david@hawaii.edu
Warren
Mathews Georgia
Tech. US warren.mathews@oit.gatech.edu
James Williams
Latin America and the
Michael
Stanton CLARA michael@inp.brFlorencio Utreras
Meeting Co-Chairs:
Shigeki Goto (Asia-Pacific), Kees 0.Neggers (Europe), and George Strawn (
II. Proceedings
1. Opening
The meeting was co-chaired by the
Continental Co-Chairs, Shigeki Goto, Asian-Pacific Co-Chair, Kees Neggers,
European Co-Chair, and George Strawn, North-American Co-Chair. The meeting organizers expressed their thanks
for sponsorship of the CCIRN dinner by Juniper Networks. Shigeki Goto expressed the regret of Kilnam
Chon that he was unable to attend this CCIRN meeting due to a recent
accident. The attendees introduced themselves.
2. Continental Reports
APAN Link Status
The
Asia-Pacific Advanced Networking (APAN) organization maintains five links to
the
The KOREN
network has a backbone of 40 Gbps to 2.4 Gbps.
The
KREONet is
a 10 Gbps network. A domestic lambda
network has been proposed to support Grid and supercomputing applications. The SuperSirenII network will be
operational in 2005-2006.
CERNET in
CSTnet is
focused on the China Academy of Sciences Institutes. They are upgrading from 2 Mbps to 150
Mbps. Their international bandwidth is 2
x 155 Mbps serving 1 million users.
The China
Next Generation Internet (CNGI) plans to provide multiple national backbones in
The CERNET2
backbone will be primarily 10 Gbps with some links at 2.5 Gbps. CNGX-IX will have links to GEANT, North
America, and APAN providing connectivity to GEANT, StarLight,
The Australian Research and Education
Network (AREN) is an alliance among Powerlink, the research and education
networking community, and Leightons. It
has been difficult to interest high capacity carriers to provide links to the
region.
GEANT connects 33 countries across
Primary joint research topics on GEANT
include security, mobility, end-to-end guarantees, and performance monitoring. Service activities include procurement,
network operations, basic services, and end-to-end Quality of Service
(QoS). Networking activities include
management of GEANT2, dissemination, support for users and user communities,
foresight study, and coordination of research and technology development (RTD)
activities.
GEANT is supporting the European very
Long Baseline Interferometry (eVBLI) experiment using primarily 10 Gbps
links. Next steps for GEANT include an
improved backbone infrastructure and support for multinational research teams
use of Grids.
CLARA
provides networking coordination in
NSF
supports international scientific connectivity.
CLARA is interested in cooperative links for cross-border fiber from
The
The US
National Science Foundation (NSF) currently supports TransPAC at $1.7 million
per year, Eurolink at $1.7 million per year, Naukanet to
StarLight, TransLight and the Global
Lambda Integrated Facility (GLIF)
StarLight
is an optical networking infrastructure in
StarLight supports 2 x OC192 links to
Amsterdam, 3 x OC192 to Canada with connectivity to Seattle, Korea, Taiwan, New
York City, and Ireland, an OC192 to London, OC48 to Tokyo (scheduled for
upgrade to OC192 in August), OC3 to Russia, OC192 to CERN, 10 GigE to Fermilab,
and several 10 GigE links to National LambdaRail (NLR). The lambdas are bandwidth concentrators. They also deliver Grid cyberinfrastructure.
Asia-Pacific Grid (APGrid) was established in July 2000
as a meeting point for Grid researchers.
It has 49 participating organizations from 15 countries. The APGrid testbed is a virtual
organization across international boundaries where participants bring their own
grass-roots approach and their own physical resources. Most contributed resources are small clusters
of equipments. Networking to support
Grid activities exists but the bandwidth is often insufficient. They hope to grow from a testbed to a
standing capability. It assumes each
institution has installed Globus Toolkit 2 (GT2). They gather and exchange trust
information. MDS is configured to build
an APGrid MDS tree. They use Globus and
additional software as needed.
Accomplishments to date include sharing resources among 26 sites and
using GT2 as common software. They have
not formalized how to use the Grid testbed and the testbed has not been
stable. They need to build an
environment to support users including a help desk, training, and Grid
management.
A Grid
working group has been established and a
JGNII is an
open testbed framework with 63 domestic access points and a maximum speed of 20
Mbps. It provides Ethernet connection
service (Layer 2) for domestic IP connections.
It also provides an OXC connection service with optical wave service
between OXC sites using 10 G services.
JGN will have a link to
EGEE has 70
partners and is the primary communication vehicle for EU Grid projects. DANTE is working on the design of the
interface between Grid middleware and the networks. The network infrastructure used is
GEANT2. They are developing a resource
allocation network resource broker and a measurements requirements
interface. CERN requirements will need
high network performance. They will
implement basic network resource reservation for premium IP service in about
month 15 of the project. Other services
will be implemented in the future with the coordination of router and switch
vendors.
A wide range of optical networking testbeds are under
development in the U.S. varying from national in scope (National LambdaRail) to
regional (CHEETAH, DRAGON) to local (I-Wire).
These testbeds respond to the need for high bandwidth data transfers and
collaborations by specific, high-requirements applications. Since they are all being architected and
constructed currently, the architects are directly collaborating to assure
commonality, compatibility, and transparency of tools, hardware, and software
to provide inter-domain services, management, security, and performance
measurement.
UltraScienceNet is being constructed
by the Department of Energy, Oak Ridge National Laboratory (ORNL) to connect
hubs close to DOE’s largest science users.
It is a sparse, lambda-switched, dedicated, channel provisioned
testbed. It provides separate funding
for research projects to support applications including high performance
protocols, control, and visualization.
It will use two dedicated lambdas in the ORNL to
CHEETAH and DRAGON are two optical
networking testbeds funded by the National Science Foundation. They provide on-demand, end-to-end, dedicated
bandwidth channels for high requirements applications to support high
throughput file transfers, multipoint collaborative computation and
interactive, remote science. CHEETAH has
participation by ORNL,
National LambdaRail (NLR) is a
national scale optical network operated and managed by a consortium of
organizations to provide facilities and services for multiple experimental and
production networks. Networks exist side-by-side on the same fiber but are physically and
operationally distinct. It provides an
experimental platform for research using optical switching and several network
layers. Initially half of the NLR
capacity will be reserved for research capable of disrupting the network. It will also deploy a high-speed Ethernet
infrastructure for WAN transport.
Point-to-point lightwaves will deploy 10 GigE with OC-192 Cisco
systems. Initially four lambdas will be
deployed, one for national switched Ethernet experimentation, one for a
national 10 GigE IP network, one for a quick-start facility for new research
projects, and one to support the Internet2 HOPI testbed.
HOPI is an Internt2
project to provide a hybrid of shared IP packet switching and dynamically
provisioned optical lambdas. Basic
service includes 1 GigE or 10 GigE unidirectional point-to-point links. It will provide a deterministic path from
CERN to
Global Lambda Infrastructure Facility
(GLIF)
GLIF is a collaboration among worldwide National Research and
Engineering Networks (NRENs), consortia, and institutions with lambdas. It was established in August 2003 as a
world-scale laboratory for application and middleware development on emerging
lambda Grids where applications rely on networks dynamically configured from
optical links. The network is becoming
the most important architectural component providing access to distributed
resources and to interconnect users.
Participants include StarLight,
1.
Governance
and policy
2.
GLIF
lambda infrastructure and lambda exchange implementations
3.
Persistent
applications
4.
Control
plane and Grid integration middleware
The GLIF 5th Workshop will be in
Latin
America and the Caribbean have two optical networking testbeds, one in
The Giga
Project in
The end-to-end (E2E) piPEs framework is a basis for the
Internet2 E2E Performance Initiative.. It enables users to identify end-to-end
network performance, locate problems, and locate the correct person to address E2E problems. It
provides measurement infrastructure components.
In Phase I it will focus on the database, web-based display engine, the
analysis engine, and performance measurement points. The regularly scheduled tests include latency
(OWAMP), bandwidth (BWTCL), and routing information (traceroute). Phase II will
provide measurement domain support, deploying a measurement infrastructure
prototype on the
The High Performance
International Internet Service (HPIIS) is ending, probably in about six
months. It will be replaced by the
International Networks Research Connections Program (INRC). The INRC provides peer review to identify highly
recommended projects. Negotiations lead
to awards and the awardees must procure infrastructure. Critical issues to address include security
and performance measurement. CCIRN is
encouraged to make a strong organizational statement on security and to foster
a focus on resources to provide security.
A work program is currently developing an Asian Observatory to extend
the Abilene Observatory.
CCIRN members discussed the
feasibility of supporting the creation of a Networking R&D journal. It was noted
that there are significant differences between the "R" and the
"D" communities, making it likely to be difficult to serve both with
the same publication. The "R" community traditionally publishes its
results and may already be adequately served by existing journals. The
"D" community typically does not publish its results. Thus, although
the need for such publishing was acknowledged, it was thought that motivating
the "D" community to publish was a larger problem than journal availability.
US journals generally use an author-funded approach to
publishing. Page charges lead to use of
publication on a server to make research or development results available. Significant publication charges could
discourage publishing by authors from disadvantaged countries. Three
-
Physics On-Line (POL):
the Archive covers a wide purview and half of physics articles appear there.
- Biomed Central: offers services in bio- and medical sciences areas. Authors prepay for submission of articles.
- Library of Science: Harold Varmus started this author-funded model. They are interested in helping to start journals in areas outside biomedicine.
A CCIRN working group could investigate what is the best model for a
networking journal. They should also
consider if there is a driving need for an online journal. Many years ago TERENA started a journal but
later experienced a serious drop in interest for having research
published.
Heather Boyles
described an Internet2 International networking project to coordinate
connectivity planning and programs for lesser-served areas such as Africa and
Warren Mathews of
Georgia Institute of Technology described NECTARnet, a concept for a high
capacity, high performance network planned initially for West Africa and
TEIN2 is an extension
of GEANT to provide additional connectivity between Europe and
EUMEDConnect is being
coordinated by DANTE to provide networking infrastructure around the
The NATO networking
panel agreed to install a regional network for the newly independent states
(NISs) of the Soviet Union, the Southern Caucuses, and Central Asia for
connection to the European NRENs and GEANT.
There is funding of $2.5 Million for three years to provide VSAT
technology with a hub in
AARnet has worked with
the Southern Cross Cable Network (SCCN) on a continuing basis. In December 2003 they announced a new service
with drop-offs of 155 Mbps dual circuits. And dual 10 Gbps circuits from
APCERT is a coalition
of members with two levels of participation.
There are currently 15 members from 12 countries. APCERT provides representation to other
regional and international bodies such as TF-CERT, EGC, and FIRST. Its goals are education, training, and
awareness. They have a traffic data
share project for the Asia-Pacific region, an observatory for network traffic
and a traffic-monitoring group.
There is a strong need
to cooperate on security since incidents are international and no one
organization knows everything. The Task
Force of Computer Security Incident Response Teams (TF-CSIRT) in Europe
provides collaboration points among the existing CERTs in
-
How to accredit CSIRTS
so they can be trusted
-
Incident Response Team
database: An incident leads to identification of a CSIRT with purview for that
address leading to a request for help from that CSIRT
-
Clearing house of
incident handling tools : Sharing
information on the tools CSIRTs use
-
Deliverables: Transfer
of knowledge from one operational person to another
-
Training for the
trainers
Internet2 is working on
network security, middleware, and Trust Federations. REN-ISAC, the Research and Education Network
Information Sharing and
-
Unscheduled outages
-
Security related
events such as DDoS attacks, virus alerts, systematic network vulnerabilities
scanning, and systematic spoofing
-
Other anomalies posing
a threat to the networks
Internet2 initiated a
middleware program in 1999 focused on the enterprise campus. It supports collaboration access to digital
resources, virtual organization conformation, and disbanding. It has resulted in the Shibboleth
authentication, transport software, and open source software being deployed in
the
The Internet2 work is
addressing issues of:
-
How do national level
trust federations trust each other?
There is an October 14-15 meeting of countries establishing national
trust federations.
-
How do you peer different country federations?
The Security at Line Speed Activity (SALSA) has a working group on
network authentication. They have a
relationship to similar activities in TERENA.
The APAN 6NOC provides
a dual-stack network. A 6GN is being
established that will be run by the 6NOC.
RNP in
CUDI has been using
IPv6 since 2000. They organized an IPv6
workshop in February, 2004 attended by
The next CCIRN meeting
is scheduled to be in conjunction with the 2005 TERENA Networking Conference
being held June 6-9 in