Find the most recent information on EU Funding activities in the field of Information and Communication Technologies (ICT) by visiting our ICT in FP7 website, which covers ICT in the 7th Framework Programme (FP7) 2007 - 2013.
IST Call 5 Preparatory Workshop on "Advanced Grid Technologies, Systems & Services"
Date: 31 January - 1 February 2005
Location: Hotel Carrefour de l'Europe, Grasmarkt 110, B-1000 Brussels
Parallel session on "Network-Centric Operating Systems"
Tuesday February 1st 2005
Session organiser: Franco Accordino
Abstract
Network-centric Grid operating systems bear the promise to become the new frontier in management of complex distributed computing systems and services. They support properties and provide functionalities that are usually addressed at middleware level to enable seamless integration and management of distributed resources while providing a uniform interface to applications and services. In a world where computing and knowledge capabilities are "escaping from the box" to pervade our everyday lives, the big challenge is to master such immense power in the same way that Operating Systems did in the past forty years for the capabilities "inside-the-box". This poses new scientific and technological challenges requiring a re-thinking of long-cherished system design and implementation principles at the crossroad of Operating Systems and Grid research.
This session provides an opportunity to identify and discuss the main research issues on network-centric Grid Operating Systems, in the light of the two alternative and complementary approaches proposed in the IST Work Programme 2005-2006: on one hand research and conceptualization of next generation Operating Systems that go beyond the conventional architectures and design principles, on the other hand, development, testing and validation of new versions of existing Operating Systems enhanced with Grid-like capabilities.
List of organisations presenting a position paper
Beihang University (China) - SIMULA Research Laboratory (Norway) - Cambridge-Cranfield High Performance Computing Faculty (UK) - Institute for Systems and Computer Engineering, Research and Development (Portugal) - Universita' di Trento (Italy) - University of Science and Technology of Lille (France) - University of Münster (Germany) - Research Centre Jülich (Germany) - Centre National d'Etudes Spatiales (France)
Agenda with speakers
- 08:30
- Keynote: A "Grid" for the Service Industry of Tomorrow
- Prof. W. Zwaenepoel (EPFL) (PDF, 1.091KB)
- Services account for an ever-increasing percentage of economic activity, and every indication is that this dramatic growth will continue or even accelerate over the next decade. Information technology will be a key enabler of this transformation. A world-wide grid of services will underlie the service industry of tomorrow. Unlike the current Internet, which is dominated by interactive access to information, the future Internet will predominantly see programmatic access to services. In this talk I will develop what I see as the key challenges that have to be met to develop the vision of an Internet of services. Many of these challenges involve scalability, namely at the user, the server, and the service level. Furthermore, complexity must be tamed by better methods for configuration, monitoring and control. A new wave of network-centric operating systems as well as Grid middleware is required to achieve virtualisation and simplification. Finally, a major non-technical challenge involves organizational restructuring necessary to integrate services in the business workflows and derive the productivity benefits for them.
- 9:45 Parallel session "Network-centric Operating Systems" starts
Invited presentations
- 09:45
- Strategic dimension of Operating System research
- Prof. K. Jeffery, CCLRC, president of ERCIM, member of NGG expert group (PDF, 180KB)
- The NGG2 Expert Group, proposed a vision and research agenda for EC funded and coordinated GRIDs research activity. The concept is that a user interacts with the GRIDs environment intelligently such that the it proposes a 'deal' to the user to satisfy her request which she can then execute - involving multiple resources of computation, information, detectors (new data collection), interactions with other users through various devices etc. - as a seemingly homogeneous 'surface' that has self-* capability across arbitrary and dynamic large numbers of nodes to give scalability, performance, reliability, access, security, privacy and other features. An important emerging problem is the capability of existing operating system platforms to support GRIDs middleware adequately, in turn to support application requirements. Europe has world-leading strength in telecommunications systems and devices, in embedded systems and in advanced software and systems - including Web Services, Semantics, Trust and GRIDs. To exploit the European strengths - and also to fulfil the 'Lisbon promise', Europe needs to create an environment for European industry-business-academia partnerships developing the GRIDs-based knowledge society to flourish. An open standards network-centric operating system (or a family of compatible systems) to support GRIDs (and thus ambient, pervasive computing) could be the key to creating the required environment.
- 10:00
- Requirements for Grid Operating Systems: Scalability, Adaptivity, Responsiveness, ...
- Prof. A. Reinefeld, Zuse Institute Berlin, member of NGG expert group (PDF, 127KB)
- Future Grids will comprise a large and dynamically changing number of nodes with different capabilities, run under different native operating systems and operated in different administrative domains with different policies and goals. It is now becoming increasingly clear that existing Grid middleware does not provide adequate support to enable easy and reliable application construction and execution in future Grid environments - partly because of the still prevailing client-server model which fails to scale to millions of nodes and which does not adapt to new or unexpected component behaviour. This leads to a concept of extending existing operating systems with a Grid Foundation Middleware that mediates between the native operating systems of various devices such as HPC, clusters, archives, PCs, PDAs, sensors, and actuators. It fills the gap between the application service middleware and the various native operating systems underneath. Among many other features, the Grid Foundation Middleware must provide true scalability to millions of nodes, adaptability to cope with newly occurring operation situations, and responsiveness to allow the negotiation of service guarantees.
- 10:15
- What mobile users want from the next generation of networked smart devices
- DW Wood, Executive Vice-president Research, Symbian Ltd. (PDF, 405KB)
- Smart mobile phones have increasingly sophisticated operating systems. The presenter, a co-founder of Symbian, has spent the last seventeen years overseeing aspects of the design and evolution of Symbian OS (and its predecessors). The talk starts by reviewing the context of the next generation of mobile smart devices, including the reasons for the very significant commercial interest in these devices. Smartphones have the potential to become, for hundreds of millions of people, the preferred mobile gateways into the digital universe - a universe that keeps on expanding and growing in importance. The main content of the talk looks at some important constraints that risk the successful take-up of such devices: issues that cause telecoms industry leaders to lose sleep. With every constraint comes opportunities. Follow-up discussion will explore connections of this complex picture to next generation Grid.
- 10:30
- Requirements and options from a mobile OS developer/user perspective
- T. Saridakis, Principal Scientist, Nokia Research Center (PDF, 5.195KB)
- The time that mobile phones were mere replacement of their fixed-line counterparts is long and irreversibly gone. Today, mobile phones have to face an ever increasing spectrum of, often conflicting, requirements, most of which could not even be conceived a few years back. Among the most demanding ones are various levels of energy awareness, different aspects of system security, and the dynamic upgrade of software at diverse system levels. These requirements will define the capabilities of mobile phones in the near future. Such capabilities will include greater operational autonomy despite a growing volume of software running on the mobile phone, and the option to obtain and use fixes for the running software (application and system level) or new features for it without interrupting the use of the phone. Also, supporting various aspects of system security would enforce the protection of copyrighted material, allow the secure operation of mobile phones in unknown networks and ensure that the hosting of applications does not compromise the consistency of the mobile device and the privacy of the data it contains. Mobile phones possessing such capabilities will be ideal platforms for a variety of applications that accompany the user everywhere and allow him to effortlessly interact with the surrounding environment and the other users in it.
- 10:45
- Operating System research challenges: reliability, power-awareness, SW engineering aspects
- Prof. W. Schröder-Preikschat, University of Erlangen, member of NGG expert group (PDF, 1535KB)
- The current generation of grids has its focus on computation intensive applications exploiting fairly heavy-weighted nodes interconnected by standard internet technology. Typically, each individual node contributes a rich set of quite powerful (hardware/software) resources to the grid. Next generation grids will extend this scenario towards fly-weighted nodes interconnected by ad-hoc network technology. A substructure of a network of (active/passive) sensors will provide the basis for sensor grids used to support e.g. high-precision forecasting of physical, chemical, as well as geological processes on earth, (distant) early warning of cataclysms, and emergency procedures. The majority of sensor nodes will offer only a very limited set of (hardware/software) resources to the grid. Every sensor node will be part of a large-scale embedded system and dedicated to run a special-purpose operating system that needs to cooperate with higher-level grid services being controlled mostly by general-purpose operating systems. One of the most critical resources of a sensor node will be energy, which calls for power-awareness not only of the hardware but in particular also of the software. This non-functional property represents an unbroken thread from lower-level operating system up to higher-level application programs, it is a cross-cutting concern of the entire system. Other aspects are reliability, especially giving sensor nodes self-protecting and -healing capabilities. The system software must come tailor-made with respect to the actual given application case and at the same time should be highly portable to other targets. Variability management will be one of the buzzwords in order to deal with the trade-off between specialization and reusability.
- 11:00
- Requirements and options from an automotive OS developer perspective
- C. Stellwag, 3Soft GmbH (PDF, 1.895KB)
- Today's cars contain a lot of software in it which is spread all over the car. Different ECUs are doing their work to assist and help the driver and the passengers. The ECUs are connected via different networks but act individually. The size of the software increases from year to year. Besides the specific requirements of the applications (like motor control or heating) some general facts have to be always considered. The most important ones are the requirements regarding low power consumption and the need to ensure highest quality at lowest HW/SW costs. Looking at the growing number of recalls we see that the industry has really problems in these areas. Network-centric operating systems can be one solution to solve the current problems by offering network-wide services and an easier interface for the user. Also grids can help to save costs, but the grid management for these systems needs further research to explore the chances of using grids within the car.
- 11:15 Discussion (45')
- 12:00 Short statement of ideas based on best position papers (5' each)
- 12:45 Discussion (15')
- 13:00 Parallel session "Network-centric Operating Systems" ends
- Wrap-up of session (PDF, 66KB)