The number of projects relying on volunteer computing and their complexity are growing fast. This distributed paradigm enables the gathering of idle resources (processing power and storage) to run large systems by providing scalable, practical and low cost platforms. The heterogeneity of the resources and their unreliable behavior call for advanced optimization methods. In particular, an efficient resource allocation is key for the systems¿ performance. This work presents a mathematical formulation and a solving approach based on a metaheuristic for the resource allocation problem. This approach is designed to deal with data-intensive applications, which must guarantee the availability of the data at all times. Moreover, a simheuristic is proposed to deal with the stochasticity of resources¿ quality. A set of computational experiments are performed to: (1) compare the performance of the metaheuristic and the simheuristic in a stochastic environment; and (2) quantify the effect of the stochasticity on the solutions.