The article highlighted the advantage of information systems in front of centralized systems in terms of ensuring the security of network infrastructure, data transmitted, and procedures used in the software environment of the complex. It was pointed out the technical simplicity of the methods of expansion of information systems and, accordingly, their scalability. A multilevel energy balance scheme and computing resources of the information system infrastructure, based on the optimization of the task graph, have been developed. The methods of estimating the optimization of the graph of tasks, which are based on the indicators of the normalized graph normalized distribution of energy, are determined. The results of mathematical modeling in comparison with statistical data for such methods of working with task graphs as calculations by hierarchical structure, graph split algorithms, methods based on algebraic theory of graphs, structuring of "Diamond Dags" type. The analysis was conducted for such types of distribution as uniform distribution, binomial distribution, geometric distribution. The results of the comparison point to a sufficiently upstart prediction accuracy at the level of mathematical modeling. In most cases, the value of the maximum deviation, presented as a relative error, between simulation results and statistical data remains within 10%, which shows the adequacy of the simulation. Nevertheless, for the four pairs of functions of the length of the normalized graph and the normalized energy distribution, the maximum deviation is greater than 10%, indicating the need for a revision of the model and further statistical studies. It is proposed to use in organizing the protection of resources of the information system of clusterization methods, which allow to construct a transparent scheme of functioning of the complex and to construct monitoring algorithms.