Market Models for Grid Computing

Most researchers in Grid computing believe that computing services will, in the future, be provided similarly to telephone, electricity and other utilities today. In that case there will be a market with different Grid companies competing and co-operating to serve customers. Companies will have to set prices to attract customers and make profits. Governments that may choose to provide GRID services will similarly have to set prices, perhaps with a different aim of maximizing utilization. This area of Grid research has been highlighted as a key area of research in the EU report on next generation Grids. The development of the Grid has come to a stage where companies are beginning to sell Grid services. For instance, the BBC and The Register recently reported that both IBM and SUN are selling Grid access by the hour. Already a market exists where two companies compete for GRID customers. It will probably not take long until we see brokers (also called middlemen or suppliers to end-users in different environments) re-selling Grid services they bought in bulk from providers. This would constitute the usual 3-tier setup of markets seen, for instance, in the electricity market, where we have generators, suppliers and user.For the companies involved, who may wish to provide GRID access as an efficient alternative to buying computers, it is important to have a market model to be able to test various pricing schemes. This will become even more evident once futures and options for Grid access are sold. Future customers of this Grid market are expected to be banks, insurance and engineering companies, games companies, as well as universities and other research organisations. Although a great deal of effort has been aimed at modelling financial markets, much less is known about modelling commodity markets. Our research will enable companies and governmental agencies to evaluate different strategies and regulations to explore different market rules in a realistic fashion. By combining our experience of e-Science, modelling markets (see VR's CV) and computer performance, we will develop a realistic market model for Grid commodities. Nowadays, modelling markets and describing human behavior in economic systems is a very lively topic in statistical physics and econophysics, so our work is expected to also impact basic research in these fields. This proposal aims to produce a high-level market model for Grid computing. In contrast to previous research we propose to investigate, not the use of economic methods for the micro-management of future Grids, but rather how a future Grid-based economy can be modelled and predicted. This will bear some similarities and analogies to current structures in place for the airline, electricity and telephone industries, but will be specifically focussed on the needs and charcteristics unique to the Grid.