Chapter 8 - 21. Congestion Management in TCP Storage Networks

发布于:2024-03-05 ⋅ 阅读:(62) ⋅ 点赞:(0)

Goals

The end-to-end flow control and congestion control mechanisms of TCP are enabled by default and no user intervention is needed. But these mechanisms, including ECN, are not invoked until needed. When these mechanisms are invoked, the following are the troubleshooting goals. TCP 的端到端流量控制和拥塞控制机制是默认启用的,用户无需干预。但这些机制(包括 ECN)在需要时才会调用。调用这些机制时,故障排除目标如下。

1. Identify the source (Culprits) and cause of congestion: The cause of congestion is most certainly the lack of capacity on a network link, inability to handle traffic burst, or bit errors. 造成拥塞的原因肯定是网络链路容量不足、无法处理突发流量或比特错误。

2. Identify the affected devices (Victims): When a network link does not have enough capacity, all flows passing through that link may be affected because their packets may be dropped or delayed. Unlike lossless networks, only the flows that pass through the congested ports are affected. Other links in the traffic path are not affected. Also, the source and destination of the affected flows can continue to send/receive traffic to/from other devices via non-congested links. 当网络链路容量不足时,通过该链路的所有流量都可能受到影响,因为它们的数据包可能会被丢弃或延迟。与无损网络不同,只有通过拥塞端口的流量会受到影响。流量路径中的其他链路不会受到影响。此外,受影响流量的源和目的地可以继续通过非拥塞链路向/从其他设备发送/接收流量。

Congestion Severities and Levels

Chapter 4, “Troubleshooting Congestion in Fibre Channel Fabrics,” defines co

本文含有隐藏内容,请 开通VIP 后查看

网站公告

今日签到

点亮在社区的每一天
去签到