PANEL
SESSIONS
Panel 1:
Charging for Quality-of-Service or Managing Service Level Agreements?
Chair: Burkhard Stiller, ETH
Zurich, TIK, Switzerland
Panelists:
Jrn Altmann, Hewlett-Packard Laboratories,
Cupertino, U.S.A.
Nevil Brownlee, University of Auckland,
New Zealand
Marcus Brunner, NEC Europe, CCRL, Heidelberg,
Germany
Georg Carle, GMD Fokus, Glone, Berlin,
Germany
Christian Rad, AT&T, U.S.A.
On the one hand, mechanisms
to provide fine-grained Quality-of-Service (QoS) in the Internet
have been designed in support of several different networking approaches.
Since QoS allows for the introduction of different service levels,
a differentiation with respect to their costs is useful. However,
a commonly agreed upon manner to apply charging functions has not
been found yet. On the other hand, the management of Service Level
Agreements (SLA) between Internet Service Providers shows quite
a static approach over recent years. Advanced mechanisms for short-term
adaptation of bandwidth being exchanged and its quality management
are coming up. The panel is open to these topics and shall discuss
related open issues: Will negotiated SLAs be sufficient for future
network interconnections to provide at least a statistical QoS to
applications? Which charging approaches are appropriate for QoS
and SLAs? Which service level and quality will be of interest for
charging anyway?
Panel 2:
Optical Internetworking: Can We manage the Integration of IP and
the Optical Layer?
Chair: Doug Rom, Telcordia
Technologies, USA
Panelists:
Bala Rajagopalan, Tellium, USA
Scott Marcus, Genuity, USA
Jim Gambony, Worldcom, USA
Bill Miniscalco, Verizon, USA
Beau Atwater, Telcordia Technologies,
USA
Dynamically configurable
optical networks are rapidly becoming a commercial reality. These
networks appear to be architecturally simpler than the multi-layer,
multi-technology networks common today, and thus may offer the promise
of simpler operation, or even self-operation. Major industry forums,
such as the IETF, OIF, the ODSI Coalition and T1X1, are developing
signaling, control and management mechanisms that would enable rapid
bandwidth provisioning and new service features within these networks,
as well as an increased level of integration between the IP and
optical layers. Direct signaling between network layers, combined
with greatly increased intelligence within optical network elements,
will carry significant implications for the distribution of network
management functionality and will impact the role and design of
carriersŐ operations support systems.
Panel 3:
Title: SNMP and/or COPS for Configuration Management?
Chair: Bert Wijnen, Lucent
Technologies, USA
Panelists:
Diana Rawlins, WorldCom, USA
David Durham, Intel Corp., USA
Steve Waldbusser, Consultant, USA
In the IETF, two working
groups are defining methods to do configuration management in the
context of policy-based management. The protocols that are intended
to be used are SNMP and COPS (COPS-PR) COPS is sort of a "new kid
on the block," while SNMP has been around for more than a decade.
Even when COPS is used, it is still assumed that SNMP will be used
for monitoring, and so SNMP will not go away. So it seems that both
protocols will probably be used in future network management, or
at least that is what some people currently think. The question
on the table is: Do we need both protocols or can we get away with
just one?
Panel 4:
Can WBEM Replace SNMP?
Chair: J.P. Martin-Flatin, AT&T
Labs Research, USA
Panelists:
Jim Davis, Sun Microsystems, USA
James Hong, POSTECH, Korea
Aiko Pras, University of Twente, The Netherlands
Juergen Schoenwaelder, TU Braunschweig,
Germany
Andrea Westerinen, Cisco, USA
SNMP-based management
is ubiquitous in the IP world. But it has exposed serious limitations
in real life, most notably its lack of scalability, its non-object-oriented
information model, and its focus on network element management and
low-level instrumentation MIBs. To date, the industry's main response
to these concerns is Web-Based Enterprise Management, a new management
architecture developed under the umbrella of a large industrial
consortium: the Distributed Management Task Force (DMTF). WBEM is
characterized by a distributed architecture, an object-oriented
information model, and the integration of network, systems, application,
service, and policy-based management. By addressing the main deficiencies
perceived in SNMP, can WBEM eventually replace it? Will customers
migrate from SNMP to WBEM? Did WBEM learn from SNMP's mistakes?
Should WBEM/CIM information modeling start with a clean slate, or
should it build on existing models? Will SNMP and WBEM co-exist
for many years?
Panel 5:
Will Pervasive Computing be Manageable?
Chair: Morris Sloman, Imperial
College, London, UK
Panelists:
Victor Bahl, Microsoft Research, USA
Gaetano Boriello, University of Washington,
USA
Richard Graveman, Telcordia, USA
Steven Shafer, Microsoft Research, USA
Computers are being integrated
into everything. Domestic appliances such as refrigerators, video
recorders or door locks could act as web servers for remote access;
your mobile communicator will seamlessly switch between Blue Tooth,
wireless LAN or UMTS to choose the most appropriate communication
link; wearable computers in smart garments will monitor your medical
well-being; your whereabouts will be tracked by integrated GPS or
by base stations; intelligent paper with integrated radio will provide
a light-weight, unbreakable, high-resolution output media for e-books
and e-newspapers and mobile robots will clean streets. Mobile computing
technology will enable users to interact with their pervasive computing
environment to obtain location dependent information, be entertained,
purchase goods or control their local environment. It is obvious
that the time and effort to configure a typical web server or workstation
will be impractical for the forecasts of 100K pervasive computers
per person which will be embedded around our future homes, offices
and leisure facilities. There is a need for self-organizing systems
that can dynamically adapt to form ad-hoc collaborating groups.
This panel will address the issues of whether current security and
management mechanisms scale to cater for millions of mobile computers
interacting with a pervasive computing environment.
|