Managing risks in an open computing environment using mean absolute deviation portfolio optimization

Users who execute their jobs in an open computing environment are exposed to risks, because open computing resources have variance in their performance. This paper proposes the mean absolute deviation portfolio optimization model to control uncertainties of open resources. The effect of the model was investigated using simulations. The simulation results show that diversified resource portfolios can satisfy users' risk requirements and maintain the stability of the open computing market only if either the risk information of the global data repository can be checked in real-time or service brokers directly record the performance history of used resources.