Chapter 8 - 3. Congestion Management in TCP Storage Networks

Bit Errors in Lossy Ethernet Networks with TCP Transport

As explained in Chapter 1, when a bit stream is exchanged over a network, some bits may be altered resulting in bit errors. These bit errors may occur because of environmental issues, such as temperature and humidity, causing interference with data transmission, faulty cables, faulty SFPs, loose cable connections, loose SFP connections, and accumulated dust on cable end points, SFPs, or patch panels. 如第 1 章所述，在网络上交换比特流时，某些比特可能会被更改，从而导致比特错误。出现这些比特错误的原因可能是环境问题，如温度和湿度对数据传输的干扰、电缆故障、SFP 故障、电缆连接松动、SFP 连接松动以及电缆端点、SFP 或配线架上的积尘。

A packet is considered corrupted when its contents are changed between a sender and a receiver. 当数据包的内容在发送方和接收方之间发生变化时，该数据包即被视为损坏。

Ethernet CRC

At layer 2 of the OSI model, the Ethernet header (Chapter 1, Figure 1-6) carries a 4-byte Frame Check Sequence (FCS) that carries the result of the cyclic redundancy check (CRC) calculation. Because the source MAC and destination MAC change at every layer 3 hop, the FCS field is re-calculated and updated in the Ethernet header before sending the frame to the next hop. An ingress Ethernet frame is found to have a CRC error when the result of the locally computed CRC calculation does not match the FCS value in the frame. 在 OSI 模型的第 2 层，以太网报头（第 1 章，图 1-6）包含一个 4 字节的帧校验序列 (FCS)，其中包含循环冗余校验 (CRC) 计算的结果。由于源 MAC 和目的 MAC 在第 3 层的每一跳都会发生变化，因此在将帧发送到下一跳之前，FCS 字段会在以太网头中重新计算和更新。当本地计算的 CRC 计算结果与帧中的 FCS 值不匹配时，就会发现入口以太网帧有