<p>To meet the diverse requirements of a multi-tiered edge cloud-based system, it is necessary to adjust resources across the system as needed. This helps maximize the system’s revenue from resource utilization and ensures that the quality of service is maintained. Due to the inefficiency of the mean demand model, stochastic demand-based algorithms are introduced when there is a significant fluctuation in real-world resource demands. However, existing algorithms are tied to fixed-level architecture and are not versatile enough for various scheduling scenarios with different numbers of levels. We have discovered an effective method to solve the general scheduling problem in a wide range of edge architectures with various levels. The goal is to allocate the appropriate number of resources in the regions while making effective use of randomness in resource demands. We perform mathematical analysis to identify the isomorphic network of subproblems with the original system architecture, and then structure their solution process in a dynamic programming manner to find an efficient algorithm. With significantly lower computational overhead, experimental results show that the revenue generated by the proposed algorithm can achieve an average of nearly 98% of the optimal solution in various settings. Therefore, it can be used as a complementary tool to the algorithms currently available in time sensitive scenarios.</p>