本文主要是介绍Chapter 8 - 15. Congestion Management in TCP Storage Networks，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
User Actions
After learning the states of queue utilization, the following are the actions that admins and operators can take while using TCP transport for storage traffic. 了解了队列利用率的状态后，管理员和操作员在使用 TCP 传输存储流量时可以采取以下措施。
1. Like any other TCP/IP network, TCP storage networks should also aim for low queue utilization. As Figure 8-12 shows, low queue utilization not only maintains high throughput and return on investments, but also minimizes the queuing delay, and maintains large headroom for traffic burst. 与其他 TCP/IP 网络一样，TCP 存储网络也应力求降低队列利用率。如图 8-12 所示，低队列利用率不仅能保持较高的吞吐量和投资回报率，还能最大限度地减少队列延迟，并为流量突发保持较大的预留空间。
2. For achieving low queue utilization, use an Active Queue Management mechanism, such as WRED, explained later. 要实现较低的队列利用率，可使用主动队列管理机制，如 WRED（稍后解释）。
3. Regardless of the type of switches in a network, congestion detection, troubleshooting, and prevention remain the same. When the queues are non-empty (low utilization, high utilization, or full utilization), the port is transmitting at full capacity. This condition is explained earlier as congestion due to over-utilization. The solution to this problem is to increase the link speed or add more capacity. 无论网络中的交换机类型如何，拥塞检测、故障排除和预防都是一样的。当队列不空（低利用率、高利用率或满利用率）时，端口正在满负荷传输。这种情况在前面已经解释过，是由于过度利用造成的拥塞。解决这一问题的办法是提高链路速度或增加容量。
4. Detecting the links that are consistently highly utilized is relatively easy. However, when the average utilization is not high, approaches like Microburst detection (explained later) and ECN counters can detect if the packets are subjected to queuing delays. If the burst is larger than can be accommodated in the queue at that time, the packets are dropped regardless of the maximum queue size. 检测持续高利用率的链路相对容易。不过，当平均利用率不高时，微突发检测（稍后解释）和
这篇关于Chapter 8 - 15. Congestion Management in TCP Storage Networks的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
