Speaker: |
Dr Mike Fisher, BT Group |
Session: |
1:
Keynote speeches – Visions of Grids and NGN |
Title of Presentation: Grids for Business: A Service Provider’s
Perspective
Revenues in the IT sector are clearly shifting from products to
services. IT services accessible over wide-area networks have the
potential to radically change the way business is conducted both locally
and globally. Grid technologies can make a major contribution to
realizing this potential. However, technical solutions suitable for
widespread commercial adoption and the open standards needed to provide
interoperability are still immature.
Current scientific deployments of Grid technology are typically special
purpose, and are complex to set up and maintain, but clearly deliver
real value to their users. Early commercial deployments also provide
benefits but most are computing clusters and make limited use of network
connectivity.
To meet the needs of business, Grids must offer users predictable price
and performance together with flexibility and control over their own
business processes. They must also allow service providers to manage
their service offerings efficiently to a wide range of customers through
the full service lifecycle.
Investment in next generation networks over the next 10 years will
amount to hundreds of billions of euros worldwide. If the technical and
interoperability challenges associated with current Grid technology can
be solved, there is the opportunity for these networks to provide a
ubiquitous infrastructure for future services.
|
|
Speaker: |
Kees Neggers, SURFnet |
Session: |
1:
Keynote speeches – Visions of Grids and NGN |
Title of Presentation: Grid networks in the research community
In his keynote Kees Neggers will present the ongoing evolution in the
research networking world towards so called hybrid networking. This
evolution started in 2001 when the first international lambda for
research networking was ordered by SURFnet between StarLight in Chicago
and NetherLight in Amsterdam. Since September 2001, Lambda pioneers meet
yearly in so called LambdaGrid Workshops. In 2003, at the third
LambdaGrid Workshop in Reykjavik, it was agreed to continue this
cooperation under the name GLIF: Global Lambda Integrated Facility.
Today, hybrid networking is rapidly moving from pioneers to mainstream
research networking, paving the way for grid computing and applications
on a global scale.
|
|
Speaker: |
Keith Knightson, Industry Canada |
Session: |
2: Setting the scene – Grid Tutorial and NGN Tutorial |
Title of Presentation: What is NGN: Architecture
This presentation will outline the basic architectural principles of NGN,
as a precursor to subsequent sessions. In particular, it will discuss
the definition of NGN, introduce the basic architectural functions and
the concept of subsystem components. The challenges of NGNs will be
described and areas for further study will be identified. A summary of
the Recommendations recently consented and under development will be
provided.
|
|
Speaker: |
Marco Carugi |
Session: |
2: Setting the scene ? Grid Tutorial and NGN Tutorial |
Title of Presentation: What is a NGN: Service Enablers
The presentation provides an introduction to the services expected to be supported in ITU-T NGN Release 1, as well as to the network capabilities from the perspective of users and service providers required for the support of these services.
Mostly based on the NGN GSI Y.2201 R1 Requirements deliverable, some details are then given about some of these service enabling capabilities, in particular IMS, Open Service Environment capabilities, Mobility, Interconnection and ?Service Enablers? (as named in Y.2201).
Driven by market expectations for new and enhanced service features in NGN, some major topics for future standardization work in the service and capabilities areas are finally introduced, as well as some initial questions about the positioning of Grid applications with respect to identified NGN capabilities.
|
|
Speaker: |
Dave Berry |
Session: |
2: Setting the scene – Grid Tutorial and NGN Tutorial |
Title of Presentation: What is a Grid?
A Grid is a system of distributed computing resources that can be
dynamically provisioned to address a variety of problems. Many examples
exist in the worlds of e-science and e-business. A Collaboration Grid is
a particularly important type of Grid that connects people in multiple
organisations to form a virtual organisation. Grids provide a
Service-Oriented Infrastructure that can underpin a Service-Oriented
Application Architecture. Grids support large-scale computing at reduced
cost while increasing operational agility. They can be seen as another
step in the provision of computing as a commodity.
|
|
Speaker: |
Richard Schlichting |
Session: |
2: Setting the scene – Grid Tutorial and NGN Tutorial |
Title of
Presentation: Grid computing: Telco perspective
While grid computing success stories from the financial, pharmaceutical,
and manufacturing sectors are abundant, grid computing at telcos is
still mostly in the formative stages. The Telco Community Group at GGF/OGF
has been working for several years to remedy this situation by
identifying opportunities for grid computing within telcos and by
promoting its use.
This talk outlines some of these opportunities, which range from simply
selling additional network bandwidth required by customer grids, to the
internal use of grids at telcos, to the offering of grid computing as a
managed service. In addition, we describe how gridsare currently used at
various telcos, and, as an example, give an overview of AT&T's long term
vision related to grid computing.
|
|
Speaker: |
Franco Travostino, Nortel |
Session: |
2: Setting the scene – Grid Tutorial and NGN Tutorial |
Title of Presentation: Make way for Grids
This talk offers a first-cut view of how Grids evolution intercepts
network evolution. The network is an all important factor in the
Grid equation. Adequate networks mean low price of admission to
Grids. Inadequate networks mean arrested development for Grids. A
key point is that Grids pose new requirements, above and beyond
fat-and-fatter pipes or low-latency pipes. While these help in
general, a Grid's footprint expands and contracts over time, as
dictated by a precise workflow, with network requirements also
changing in space and time over the lifecycle of a Grid. Therefore,
premium features such as bandwidth on demand and advance
reservations resulted in greater confidence in a Grid's operation,
yet without resorting to some wasteful, static peak allocation.
Ideally, a Grid will harness an agile network much the same way it
drives allocation of CPU and Data resources (multi-resource
orchestration). Experimentation indicates that IP services and
optical services have their own strengths and weaknesses in meeting
Grid's requirements. The talk closes with a brief review
network-related activities underway at the OGF.
|
|
Speaker: |
Dominique Verchere, Alcatel |
Session: |
3: Management, Control and Interoperability Issues |
Title of Presentation: “Grids optimized Network Control Plane”
Operational Networks are shared between different user-applications.
Critical applications such as Grid Computing, Networked remote Storage
or other Networked remote Applications (e.g. Physics, Financial
simulations) require complex connectivity services to interconnect
different types of resources (e.g. Computational, Storage, or specific
instruments). These applications specified by OGF define the class of
«Grid applications».
The Grid applications have sporadic but QoS-constrained connectivity
needs, such as high bandwidth for a massive data transfer, low latency
for real time computing result remote visualization. To achieve such a
fast connection triggering with QoS guarantees, applications should be
able to communicate directly their own QoS requirements (bandwidth,
acceptable delays and packet loss) end-to-end to the network control
functions. The recent advances in transport switching systems enable to
deliver configured, large capacity and determinist QoS connections (e.g.
low BER and no jitter). However, Grid applications are currently
overlaid on the transport networks, precluding the automation on a per
connection basis of their QoS requirements to the network control.
Interfacing network control with application control is necessary to
overcome this limitation of the overlay model. With the continual
achievements of the standard organizations such as IETF for GMPLS
control protocols, ITU-T for the ASON recommendations, and OIF for the
UNI/NNI agreements, the network control functions are numerous and
stable. GMPLS protocols can be deployed for different switching
technologies including IP/MPLS, Ethernet, SONET/SDH, and DWDM and their
combinations. GMPLS provides already a strong level of resource
virtualization at the network level. The next step is to design richer
control network capabilities to enable the Grid application services to
request more directly and automatically the correct amount of QoS needed
and during the time a Grid application session requires the connectivity
services. Moreover, further optimization can be achieved if the
selection of the localization of end resources used by applications
(such as CPU or storage) takes into account the network resources
(available bandwidth) in order to jointly manage the end resource and
network capacity for a global grid service responding to the application
needs. This challenge can be achieved by providing a consistent level of
abstraction and virtualization of resources for both the Grid
application and the Network resources to enable the co-selection. The
signaling engines should allow the combined reservation of
computational, storage and network resources. This extended signaling
approach addresses the co-allocation and cross optimization of Network
resources and Grid resources (computer, storage, instrument/sensor). The
associated scheduling functions perform the orchestration of the
different types of the resources (network, computer, storage) spanning
multiple locations and crossing different network domains. These new
capabilities impose to extend existing protocols of the service plane
(i.e. application, middleware) and the transport network plane or to
specify new protocols. They have to be designed by considering:
heterogeneous resources, advance resource signaling requirements, the
scalability related to the number of sessions, combination of resource
types, and the complexity of the connections.
From this level of consistency (i.e. harmony), Partner Grids will be
deployed to allow firms to interact/transact with their customers, their
suppliers and their industrial partners. The Virtual Organization
concepts will become mature to be implemented offering the industrial
institutions to have transparent access to shared Grid resources
(computers, software programs, data and specific instrument resources)
over shared transport networks.
|
|
Speaker: |
Michael Haley,
IBM |
Session: |
3: Management, Control and Interoperability Issues |
Title
of Presentation: Implications for Next Gen Networks and Grid Computing to support IPTV and IMS Infrastructures
This presentation will briefly review several areas of IP-based
services driving next generation networks. IPTV/VOD and IMS have
some standardization but are early in deployments and pilots,
respectively.
IPTV/VOD success will be critically hinged on view Quality of
Experience.
Newly
emerging Web 2.0, focusing on social networks and collaboration, is
cited as a potentially large but as yet unknown load on incumbent
telco networks and communty aggregators (eg, Google). Leveraging
grid and virtualization in these infrastructures can be successful,
but so far few deployments exist. We offer the concept of loosely
coupled vs. tightly coupled elements and loading attributes in
target NGN architecture which may guide successful choice of when to
use grid/virtualization for IP services, which may ultimately be
standardized by ITU-T or other body.
|
|
Speaker: |
Huilan Lu, Lucent Technologies |
Session: |
4: QoS, Performance and Security Aspects |
Title of Presentation: QoS for Next Generation Networks
Next Generation Networks (NGN) are characterized, among other things, by
the prevalent use of a common packet transport for delivering a wide
range of applications, from non-real-time to real-time, from single
medium to multimedia. The advent of such general multi-service networks
marks a major paradigm shift from today’s specialized networks with
optimized performance for respective applications and gives rise to the
need for supporting quality of service as dictated by various
applications dynamically. This presentation will give an overview of the
Q.4/13 activities on QoS support for NGN. In particular, it will
highlight the emerging standard approach to dynamic, application-driven
resource management that is known as the Resource and Admission Control
Functions (RACF). Applicable to all network-controlled applications, the
RACF can be used edge-to-edge or end-to-end, and be realized in various
ways to support different business models.
|
|
Speaker: |
Martin Dolly, AT&T Labs |
Session: |
4: QoS, Performance and Security Aspects |
Title of Presentation: NGN Security
This talk describes the challenges of NGN security and reports on the
status of the relevant work in ITU-T, particularly work in SG13, which
is the lead Study Group for NGN. The responsible Rapporteur group,
Q15/13, has defined “trust models” to which the security requirements
are to be applied. In addition, defines the security, and
authentication/authorization relationships that need to be addressed in
an NGN deployment.
|
|
Speaker: |
Michael Fehse, T-Systems |
Session: |
4: QoS, Performance and Security Aspects |
Title of Presentation: Steering via SLA
In order to facilitate industrial acceptance of grid technologies, it is
essential that they build on foundations of normal business activities.
One of these foundations is the SLA, a contract at the business level
that is not grid specific and lies in the domain of expertise of
lawyers.
SLAs are always bilateral and contain information about the partners,
period of service, obligations on the partners and payment informations.
SLAs are always end-to-end including all necessary components.
The translation of the requirement to use these SLAs into working
prototypes is the function of the “proprietary smart bit” (PSB) of the
service provider. The PSB also juggles the dynamically changing
collection of SLAs on the more likely static collection of service
objects.
To ensure the possibility of an automated offer/acceptance process and
interoperability between different service providers it is mandatory to
standardize the semantics of SLA objectives and parameters. T-Systems
and the other partners of the NextGrid consortium are therefore
evaluating proposals to standardization bodies.
|
|
Speaker: |
Niranth
Amogh
Huawei |
Session: |
5: Future Trends and Issues |
Title of Presentation: Self Adaptive Overlay Network Innovating NGN Architecture
Overlay networking and P2P concepts prove to be powerful in enabling
pervasive adoption of services. It is important to learn from these
emerging concepts and apply the features to the evolution of NGN and
Grid infrastructure. This presentation provides a perspective for future
NGN and Grid based on Self Adaptive Overlays and attempts to broaden the
vision of Overlay for Future Generation Networks. |
|
Speaker: |
Prof. Piet Demeester
Department of Information Technology
Broadband Communication Networks (IBCN)
Ghent University - IBBT |
Session: |
5: Future Trends and Issues |
Title of Presentation: Advanced Grid applications
This presentation will highlight some recent application domains where
grid computing may play a major role. A first application area is the
production of Media content (e.g. in broadcaster environments) where the
major challenge is the distributed storage and processing of video based
content. A second application area is the consumer environment where
grids may offer an alternative to classical desktop computing. A last
application area is wireless thin client computing where the terminal
has limited capabilities (processing, storage and power). |
|