An application consists of pair of processes which communicate over the network (client-server pair). These processes send and receive messages, into and from the network through a software interface called socket. Considering the analogy presented in the book "Computer Networking: Top Down Approach". There is a house that wants to communicate with other house. Here, house is analogous to a process, and door to a socket. Sending process assumes that there is a infrastructure on the other side of the door that will transport the data to the destination. Once the message is arrived on the other side, it passes through receiver's door (socket) into the house (process). This illustration from the same book can help you: Sockets are part of transport layer, which provides logical communication to applications. This means that from application's point of view both hosts are directly connected to each other, even though there are numerous routers and/or switches between them. Thus a socket is not a connection itself, it's the end point of the connection. Transport layer protocols are implemented only on hosts, and not on intermediate routers. Ports provide means of internal addressing to a machine. The primary purpose it to allow multiple processes to send and receive data over the network without interfering with other processes (their data). All sockets are provided with a port number. When a segment arrives to a host, the transport layer examines the destination port number of the segment. It then forwards the segment to the corresponding socket. This job of delivering the data in a transport layer segment to the correct socket is called de-multiplexing. The segment's data is then forwarded to the process attached to the socket.

终于. .太多的人将套接字概念连接到双端点通信，主要是在TCP/IP协议上。但是:

NO - Socket is not related to a two-endpoint communication. It's the local endpoint, which can or cannot be connected on the other side (Think about a server socket listening for incoming connection) NO - Socket it's not strictly related to TCP/IP. It is defined with a protcol, which can be TCP/IP, but can be anything else. For example you can have socket that communicates over files. You can also implement a new protocol yourself to have a communication over USB lamp which sends data by flashing: that would still be a socket from the application point of view.

关于端口概念，你在其他答案上读到的是正确的。Port通常是TCP或UDP数据包中的数字值(2字节，0-65535)。我要强调的是，TCP或UPD不一定用于IP之上。所以:

不-说端口是TCP/IP或UDP/IP的一部分是不对的。它是TCP或UDP或任何其他定义和使用它的协议的一部分。IP不知道什么是端口。

尽可能简单地说，套接字和端口之间没有物理区别，例如PATA和SATA之间的区别。它们只是一些读写网卡的软件。

A port is essentially a public socket, some of which are well-known/well-accepted, the usual example being 80, dedicated to HTTP. Anyone who wants to exchange traffic using a certain protocol, HTTP in this instance, canonically goes to port 80. Of course, 80 is not physically dedicated to HTTP (it's not physically anything, it's just a number, a logical value), and could be used on some particular machine for some other protocol ad libitum, as long as those attempting to connect know which protocol (which could be quite private) to use.

套接字本质上是一个私有端口，为连接方知道但其他人不一定知道的特定目的而建立。底层传输层通常是TCP或UDP，但也不一定非得如此。最基本的特征是两端都知道发生了什么，不管发生了什么。

这里的关键是，当在某个端口上接收到连接请求时，应答握手包括有关为服务特定请求者而创建的套接字的信息。后续通信通过该(私有)套接字连接进行，而不是服务继续侦听连接请求的公共端口连接。

套接字基本上是网络通信的端点，至少由一个ip地址和一个端口组成。在Java/ c#中，套接字是双向连接一侧的高级实现。

还有Java教程中的一个(非规范的)定义。

首先，我认为我们应该先了解一下什么构成了从a到B的数据包。

网络的一个常见定义是使用OSI模型，该模型根据目的将网络分离为许多层。有几个重要的，我们将在这里介绍:

The data link layer. This layer is responsible for getting packets of data from one network device to another and is just above the layer that actually does the transmitting. It talks about MAC addresses and knows how to find hosts based on their MAC (hardware) address, but nothing more. The network layer is the layer that allows you to transport data across machines and over physical boundaries, such as physical devices. The network layer must essentially support an additional address based mechanism which relates somehow to the physical address; enter the Internet Protocol (IPv4). An IP address can get your packet from A to B over the internet, but knows nothing about how to traverse individual hops. This is handled by the layer above in accordance with routing information. The transport layer. This layer is responsible for defining the way information gets from A to B and any restrictions, checks or errors on that behaviour. For example, TCP adds additional information to a packet such that it is possible to deduce if packets have been lost.

TCP包含了端口的概念。这些实际上是Internet套接字(AF_INET)可以绑定到的同一IP地址上的不同数据端点。

UDP和其他传输层协议也是如此。从技术上讲，它们不需要以端口为特色，但这些端口确实为上层的多个应用程序提供了一种使用同一台计算机接收(实际上是发出)传出连接的方法。

这就把我们带到了TCP或UDP连接的解剖。每个都有一个源端口和地址，以及一个目标端口和地址。这样，在任何给定的会话中，目标应用程序都可以从源进行响应和接收。

因此，端口本质上是一种规范强制的方式，允许多个并发连接共享相同的地址。

现在，我们需要看看如何从应用程序的角度与外部世界通信。要做到这一点，你需要询问你的操作系统，因为大多数操作系统都支持伯克利套接字的方式来做事情，我们看到我们可以从应用程序创建包含端口的套接字，像这样:

int fd = socket(AF_INET, SOCK_STREAM, 0); // tcp socket
int fd = socket(AF_INET, SOCK_DGRAM, 0); // udp socket
// later we bind...

太棒了!因此，在sockaddr结构中，我们将指定端口，bam!完成工作!嗯，差不多了，除了:

int fd = socket(AF_UNIX, SOCK_STREAM, 0);

也是可能的。啊，这可真是个麻烦!

好吧，实际上并没有。我们所需要做的就是想出一些合适的定义:

internet套接字是IP地址、协议及其相关端口号的组合，服务可以在其上提供数据。tcp端口80,stackoverflow.com是一个互联网套接字。 unix套接字是一个在文件系统中表示的IPC端点，例如/var/run/database.sock。 套接字API是一种请求应用程序能够向套接字读写数据的方法。

瞧!这样就把事情整理好了。在我们的方案中，

端口是一个数字标识符，作为传输层协议的一部分，标识应该响应给定请求的服务号。

因此，端口实际上是形成互联网套接字所需的一个子集。不幸的是，“套接字”这个词的意思恰好被应用到几个不同的概念中。所以我衷心建议你为你的下一个项目命名套接字，只是为了增加混乱;)

似乎有很多答案将socket等同于2台PC之间的连接。我认为这是完全错误的。套接字一直是一台PC上的端点，可能连接也可能不连接-当然我们都在某些时候使用侦听器或UDP套接字*。重要的部分是它是可寻址的和活跃的。向1.1.1.1:1234发送消息不太可能起作用，因为没有为该端点定义套接字。

套接字是特定于协议的-因此，TCP/IP和UDP/IP都使用* (ipaddress:port)的唯一性实现与eg不同。， IPX (Network, Node, and…嗯哼，套接字——但是一个不同的套接字是指一般的“套接字”术语。IPX套接字号相当于IP端口)。但是，它们都提供了唯一的可寻址端点。

由于IP已成为主导协议，端口(在网络术语中)已成为UDP或TCP端口号的同义词——后者是套接字地址的一部分。

UDP is connection-less - meaning no virtual circuit between the 2 endpoints is ever created. However, we still refer to UDP sockets as the endpoint. The API functions make it clear that both are just different type of sockets - SOCK_DGRAM is UDP (just sending a message) and SOCK_STREAM is TCP (creating a virtual circuit). Technically, the IP header holds the IP Address, and the protocol on top of IP (UDP or TCP) holds the port number. This makes it possible to have other protocols (eg. ICMP that have no port numbers, but do have IP addressing information).