----------------------------------------------------
Abstracts - Engineering Workshop
----------------------------------------------------


Title: The Abilene Observatory: Experiences, Implementation, and Projects
Chris Robb, Indiana Univ.

The Abilene Observatory project was started as an
experimental "playground" for researchers to conduct experiments on the
Abilene network itself. Begun with the Abilene upgrade in late 2002,
the project is over a year old. There are a number of research and
measurement projects that have begun within this time period. This talk
will survey those projects and give detail on what sorts of challenges
engineers came up against.

Title: Status of FAST TCP and other TCP alternatives
John M. Hicks, Indiana Univ.   
 
Grid technologies are maturing, promising to drive the next generation of
e-science.  A number of e-science applications involve moving large amounts
of data over a wide geographic region.  The physics grid community, for
example, envisions the need to move datasets in the petabyte to exabyte range
around the globe.  Standard TCP stacks, such as Reno and Vegas, are typically
used to implement reliable data transfers.  Current TCP implementations are
too inadequate to take advantage of today's high speed networks.  There are a
number of research projects underway examining the problems with TCP as it
relates to the needs of today's high performance applications.  This talk
will concentrate on a brief status overview of FAST TCP, Scalable TCP, and
High speed TCP, addressing implementation, availability, and brief application
examples.


Title: Case Studies on Intra-Domain Routing Instability
Zhang Shu, CRL

In this presentation, we introduce several case studies on the intra-domain
routing instability. We first present two statistical results of OSPF link
state updates that we collected from the network of APAN Tokyo-XP and the
WIDE Internet. We show that although most network operators do not notice,
unexpected routing instability can occur and sometimes it can occur quite
frequently. We also give some reasons of the observed instability. We
briefly introduce an ongoing cooperation work that we are conducting with
the Internet2 community to collect and analyze IS-IS data on the Abilene
network at the end of this presentation.



Title: PlanetLab: A Blueprint for Introducing Disruptive Technology into the Internet
Scott Karlin, Princeton University
 
PlanetLab is an experimental global network of computers, designed to allow
researchers to develop and test powerful new types of software that are not
confined to a single computer, treating the global network as one large,
widely distributed computer.
A new class of geographically distributed network services is emerging, and
the most effective way to design, evaluate, and deploy these services is by
using an overlay-based testbed.  Unlike conventional network testbeds, this
new approach supports researchers who want to develop new services and
clients who want to use them.
This talk describes this high-level vision, and reports the status and plan
for the realization of the vision in PlanetLab.


Title: e-VLBI over TransPAC
Masaki Hirabaru, David Lapsley and Yasuhiro Koyama, CRL

Very-Long-Baseline Interferometry (VLBI) has been used by radio
astronomers for the last 30 years as one of the most powerful techniques for
studying objects in the universe at ultra-high resolution and measuring earth
motions with ultra-high accuracy. The transmission of astronomical VLBI data
via high-performance networks is dubbed 粗-VLBI?・The potential advantages
for scientific productivity and technical operations of e-VLBI over
traditional VLBI are: faster turnaround of results, higher sensitivity of
observations, lower costs, and quick diagnostics and tests.
Recently, researchers at Communications Research Laboratory in Japan
and Haystack Observatory in the United States have performed a series of e-VLBI
experiments across APAN/Transpac and Internet2. These experiments range from
network performance and characterization, to the transfer of VLBI data in near
real-time. They have set a solid foundation for the future "operationalization"
of e-VLBI, as well as for future high bandwidth e-VLBI experiments.
In this presentation, we will discuss the current state of e-VLBI
and the results of these recent experiments. We will also discuss planned
"last mile" connectivity upgrades as well as the future of our experimental
program. Current and future network requirements will also be discussed.
 

Title: Engineering Solution for TCP over Wireless Links
Fu Cheng Peng, Nanyang Technological University

Diverse links such as wireless links, satellite links and ADSL links are
being widely introduced in current Internet, unlike wired links, these
heterogeneous links are causing significant performance degradation of
network performance, particularly on TCP, in which random loss are
always misinterpreted as congestion loss and thus erroneous protocol
congestion control are adopted.
TCP Veno (IEEE JSAC Feb., 2003) is proposed to eliminate TCP's suffering
in wireless environments. A key ingredient of Veno is that it monitors
the network congestion level and uses that information to decide whether
packet losses are likely to be due to congestion or random bit errors.
The real experiments and live Internet measurements validated Veno's
throughput improvement and its harmonious co-existence with legacy TCP
connections.
In this talk, we showed how to easily deploy Veno TCP into current
proxies-based environment within intranet. We do not make any
modification or installation on the terminal users because Veno TCP only
refines the sender-side protocol of Reno without changing the
receiver-side protocol stack. Thereafter, all wireless users within this
intranet, when accessing Internet, will not longer suffer performance
degradation, their behavior will be same as those wired users.
Furthermore, we extend this deployment solution to the multicast case
over hybrid wired/wireless links. Live measurements are conducted to
validate it. We hope this engineering solution not only enhances the
understanding of Veno TCP in Internet, but also illuminates some
philosophies beyond Veno.

 
Title of the speech: Low Cost, Long Haul Gigabit Ethernet
Alan Cowie, AARNet

An AARNet member site in rural NSW required a bandwidth upgrade for their
600km tail into AARNet's Sydney International GigaPoP. The offerings from
the established Carriers in this market were expensive and limited in
scope. An opportunity arose for AARNet to work with in partnership with a
State Utility to light-up excess dark fibre in the power transmission
network. Low cost switched ethernet equipment was chosen in preference to
traditional optical equipment.
This talk will cover the design, implementation and operation of  the network.


Title: Global Observatory for Advanced Network Operation
Yoshinori Kitatsuji, Jin Tanaka and Kazunori Konishi, KDDI

With the growth of APAN, eXchange Points (XPs) require advanced
operation for supporting multiple services as well as high performance
experiments such as Grid Computing. It's getting more difficult for
ordinary operators to deal with new services and technologies, because
they must utilize the network resource as much as possible while the
conventional services should be maintained.
This talk includes (1) comparison between APAN-JP NOC pages and Abilene
NOC/Observatory, providing details to NOC operators and users for
better services, (2) proposal of advanced services enabling resource
sharing and its scalable operations that we have learned from multiple
experiments & demonstrations with high TCP performance, and (3) discussion
on Global Observatory infrastructure and its framework to enable more
high performance experiments and demonstrations over High Performance
Research & Education Networks.


Title of the speech:  The 21st Century Intelligent Network
Tony Li, Carl DeSousa 

Abstract: Today's network architectures will impact the economics and capabilities of future applications. Profitability is more than ever a prerequisite for new applications to succeed.  Viable but bandwidth intensive applications will require a network infrastructure with intelligence.  Intelligent high speed packet delivery must perform across hybrid lambda and electrically switched topologies.  The challenges include the use of traffic engineering and intelligent fault recovery to step into the next era of global network availability.