Tcp – In TCP Congestion Avoidance Phase, are cwnd and CongWin the same thing

In short: Flow control makes sure that the receiver is never overloaded with more data it can handle whereas congestion control is used to avoid congesting the network between sender and receiver.

Flow control: each ACK the receiver sends to the sender includes the current size of the receive window, which states how many more bytes fit into the receive buffer. This procedure is called the sliding window protocol.

Congestion control: after TCP was introduced, it was noticed, that too much traffic resulted in a so-called congestion of the network: not all packages could be delivered when the network bandwidth was exceeded. The TCP implementation reacts to missing ACKs by sending these packets again, which however only makes the situation worse, as the network becomes even more congested.

The answer to this problem is the congestion control. It is quite complex, but basically what happens is, that as soon as a message can't be delivered (when a message isn't acknowledged after a certain timeout, it is considered that the message couldn't be delivered), it is assumed that the package was lost because of a congestion in the network, and the sending rate is reduced. This is also implemented using a sliding window, the congestion window. The actual algorithm is more complex, it nowadays includes the slow-start, congestion avoidance, fast retransmit, and fast recovery algorithms.

At any given moment, a TCP implementation is allowed to send at most min(receive window, congestion window) unacknowledged packets to respect both the flow control and the congestion control.

1. A host indicates the largest TCP segment size (local MTU - (IP overhead + TCP overhead)) it can receive by the MSS. The MTU is the largest IP packet the underlying L2 transport can send/receive. Usually, they are directly related.
2. The receive window is another TCP connection parameter not directly related to MSS or MTU. It's rather about path throughput and latency.
3. Likely there are slower hops in the path than either receiver or sender can see. They must not be overrun. Often both receiver and sender have fast local connections (at least 100 Mbit/s) while they are connected over much slower Internet connections (say 10 Mbit/s).