Development of a catallaxie paradigm for the decentralized use of dynamic application networks
The research project CATNETS investigates the new possibilities of grid computing.
Basics - Grid Computing
In grid computing (based on the English term power grid), users can access computers, databases or connected devices “at the push of a button”. In Germany, for example, the Federal Ministry of Education and Research (BMBF) has been funding the development of processes and services for scientists as part of its e-Science initiative since 2005. The aim is to develop virtual knowledge environments in which users can dynamically access comprehensive databases, visualisations and scientific information of all kinds on the Internet, regardless of the computers and programs available on site.
Grid computing should thus provide an alternative to the procurement of mainframe computers in science and industry. In fact, worldwide only about 5% of the storage and computing capacity available in normal office computers is permanently used. By networking via the Internet, these capacities could be interconnected to form virtual supercomputers or huge databases. Similar to the power grid, where electricity comes out of a wall socket, computer users will then be able to obtain computing time or storage capacity somewhere on the Internet without having to worry about the complex network design behind their wall socket. This means a significant reduction in investment costs for companies that only buy and pay for computing power when needed.
So how does such a grid work? Simplified: All distributed resources are connected via grid software. First of all, there are tiny software sensors that are located in the individual resources, monitor them and, if necessary, provide them with jobs. These sensors – and thus the resources themselves – are connected to a master or resource broker, which stores all information of all available resources in a catalogue and updates it permanently. It also receives the incoming computing orders and forwards them to a suitable, available computer in the Grid on the basis of certain agreements with the customers (Service Level Agreements, SLAs). Other components of the grid software architecture ensure that the grid is equipped with a security system so that only authorized users are accepted by the grid.
Other services enable access to the user’s applications and data, which may be distributed somewhere in the grid, or they ensure the desired delivery of the results after successful work. The interfaces between the various components are standardized so that they can be combined or exchanged as required. The Grid concept has now reached such a stage that Grid projects are currently emerging in all parts of the world with the aim of establishing reliable production platforms (so-called global grids and enterprise grids) for scientists, engineers and commercial users.
Innovation in the Project
The research project CATNETS, launched on the initiative of the University of Bayreuth, shows alternatively a new, scalable possibility for resource allocation based on concepts of economic self-organisation. It is funded by the European Union in the “Future and Emerging Technologies” program. The research project is intended to show the differences between allocation procedures in the case of increasing amounts of resources, geographical network expansion or fluctuating reliability of connections.
For the first time, the project also conceived a two-tier market, which divides grid computing into a service and resource market. This model takes into account the dynamic allocation of components at application and infrastructure level. As an analysis tool for this purpose, a key figure framework was developed that transfers the technically ascertainable metrics into economically relevant decision parameters. This allows costs and revenues for the use of different allocation rules to be displayed and evaluated with only one superordinate key figure.
CATNETS has a financing volume of 1.4 million euros, distributed over a three-year project duration from 2004 to 2007. 6 European institutions from the fields of business informatics, economics and computer science are researching the common research question under the leadership of the Bayreuth Chair of Business Informatics, Professor Dr. Torsten Eymann. Besides Bayreuth, the consortium is formed by the Universität Karlsruhe (TH), the Technical University of Catalonia in Barcelona (Spain), the Universities of Ancona (Italy), Cardiff (Great Britain) and the Istituto Trentino di Cultura in Trento (Italy).
Further information: www.cordis.europa.eu