Research on Task Scheduling Optimization Model in Stream-Batch Integrated Computing Architecture Based on Stochastic Dynamic Factors

As the scale of distributed computing systems grows, task scheduling in stream-batch integrated computing architecture has become increasingly complex, especially when it needs to address resource fluctuations and uncertainties in task execution time. Traditional task scheduling methods perform poorly in dynamic environments and struggle to effectively handle these changes. This study proposes a task scheduling optimization model based on Mixed-Integer Linear Programming (MILP) and dynamic scheduling strategies. The model combines heuristic greedy strategies with local search algorithms to minimize task completion time and improve resource utilization, while addressing dynamic resource changes and uncertainties in task execution time. Experimental results demonstrate that this optimization model significantly reduces task completion time and improves resource utilization compared to traditional scheduling methods, with strong dynamic adaptability. Especially in cases of significant resource fluctuations and task delays, it maintains efficient scheduling performance.