As the bandwidth allocated to the unit buffer continues to decrease, contention for burst flows occurs instantaneously, posing significant challenges to both application performance and network stability. Current buffer management (BM) mechanisms focus primarily on ensuring fairness in the allocation of buffer resources across different ports, but often fail to effectively deal with traffic contention, either within the same port or across different ports. These mechanisms struggle to effectively absorb bursty traffic, increasing the urgency to optimize buffer utilization. Consequently, it is critical to maximize the efficiency of buffer resource utilization while maintaining isolation, especially in scenarios involving incast or contention. To address this issue, we introduce TAB, an innovative traffic-aware BM mechanism that provides a flexible and easy-to-deploy solution. TAB autonomously adjusts its parameter settings in response to traffic variations, thereby minimizing unnecessary packet loss and improving overall performance. It is also capable of adapting to different network environments and application requirements. We demonstrate the benefits of TAB for congestion control algorithms (CCAs), such as HPCC, through extensive large-scale simulations. The results show that TAB significantly reduces the 99th-percentile latency of foreground flows on both the same and different ports handling background flows by up to 8.8% and 8.5% respectively in Web Search, while having a negligible impact on the performance of background flows.