Entering the B5G era poses unprecedented challenges for data centers, characterized by burgeoning data volumes and the proliferation of edge computing, leading to uneven resource allocation. The sharp increase in energy consumption poses a stern test to environmental sustainability, compounded by rising maintenance costs due to technological complexity. Addressing these challenges necessitates data centers to innovate in architectural design, energy efficiency management, and cost control to support future demands for ultra-high-speed, low-latency, and highly intelligent communication systems. In response, this paper proposes the Threshold Adaptive (TA) dynamic migration strategy based on health and resource utilization thresholds in the baseband pool. This strategy adjusts dynamically to mitigate energy consumption, optimize resource utilization, and reduce migration frequency and duration. Specifically, it includes a Kalman filter-based trigger migration strategy, TA-VM for optimal virtual machine selection pre-migration, and TA-Target for post-migration server selection based on CPU and memory utilization. These innovative approaches aim to enhance data center operational efficiency and promote green sustainability amidst the rapid evolution of cloud and edge computing requirements.