C++ tcp server example
TCP Server Example in C++
The two protocols that make up TCP/IP specify how computers can connect with one another via the internet. Servers that interact with clients using TCP/IP are known as TCP/IP servers.
TCP
Between two communication partners in a computer network, the Transmission Control Protocol (TCP) is a protocol that ensures dependable end-to-end communications. It is in charge of dividing up the data that needs to be sent in smaller packages, combining them in the right order, and preventing/fixing package loss and network congestion. Robert E. Kahn and Vinton G. Cerf created TCP in 1973, and it can be viewed as a substitute for the transport layer in the OSI model.
IP
The Internet Protocol (IP) is in charge of routing, such as determining the best route via the network to the intended recipient. Following a predetermined pattern, the IP distributes participants in the network addresses (the IP addresses) (see IPv4 vs. IPv6). The devices connected to the network can communicate with one another by sending information to the appropriate IP address. The Institute of Electrical and Electronics Engineers released the IP protocol's initial publication in 1974. (IEEE).
History
TCP and IP were combined and standardized in 1981.
Usage
TCP/IP can be used to deliver all kinds of data across a network from one device to another and back. Popular applications are:
- Internet browsing
- Remote login
- Remote server access
- Network file transfer
Working of TCP
Two conditions must be met for a TCP connection to be successful: the target port must be open, and each communication partner must have a distinct IP address. The port enables linking a particular TCP/IP connection with the associated server-side application.
Connection Establishment
The following is how the link is established:
The client sends the server a SYN package (SYN = synchronise) in the first step along with a random sequence number. The sequence number is a crucial tool for making sure that packages are transmitted completely and in the right order.
The server accepts the connection after receiving the SYN packet by responding with a SYN-ACK package (ACK=acknowledgement), which contains the sequence number raised by one to signal the synchronisation request was successfully received. It also contains a second (again random) sequence number to signify the desire for a reciprocal relationship.
The SYN-ACK packet is delivered to the client. It then sends back the so-called ACK package, which includes both the server's sequence number and an additional one of its own. Keep in mind that the opposite side increased both sequence numbers once following this package to signify a successful connection formation.
The connection is made after the server receives the ACK package. Up to now that is till today, no actual payload data has been transmitted.
Payload Exchange
The client sends the server the fourth package, which is the first to contain payload data (typically containing the HTTP request). The two sequential numbers remain the same. Assume that the payload is 104 bytes long.
The first package with the payload is returned by the server after receiving the package. He modifies the client's sequence number in the package to 105 (+104 bytes) to reflect how much data has already been received. The sequence number of the server doesn't change. Assume the package is 1200 bytes in size.
The client receives the package with it’s sequence number set to 105. It is now knows the server has already received 105 bytes of data. It builds the next package with payload data and changes the server’s sequence number to 1201 (+1200 bytes) and sends it.
The package is delivered to the client with the sequence of the number set to 105. It is aware that 105 bytes of data have already been sent to the server. The next package is constructed with the payload data, and it is sent once the server's sequence number is changed to 1201 (+1200 bytes).
If the client is solely interested in receiving data and does not need to send any more, its sequence number will not change while the server's sequence number rises with each package.
This procedure is carried out up until the client receives the server's entire payload of data.
Connection Tear-down
After acknowledging the final item is the server transmitted, the client chooses to cut off the connection. As a result, it transmits the empty FIN package (the client's sequence number stays at 105 while the server's sequence number is equal to the beginning number plus the total amount of bytes the client got). Assume that 10004 is the sequence number of the previous server.
The server accepts the termination and sends the client a new package with an updated sequence number (to 106). The sequence number of the server doesn't change.
Finally, the client completes the connection by sending the last package, which has its own sequence number unchanged and the server's sequence number raised by one (10005).
IP Address
The IP address is a special address within a computer network that enables transmittion of the data to the intended recipient, as it was already established. They are attached to a particular gadget rather than a particular place in the physical world. The protocol establishes is how the source of the data and destination of the data are described; the IP (internet protocal) defines a package structure that condenses the payload data. It adds the "IP header" to data packages to distinguish this meta data from the actual payload. IP datagram is another name for an IP header plus payload.
A host in a TCP/IP computer network is identified by a 32-bit unique number. Any network-connected device, such as a computer, phone, or printer, can serve as a host. The dotted-decimal representation of the IP address is as follows:
192.168.123.32
The IP address can be distinguished into two separate parts:
The network component outlining the package's final destination the host component, which describes the network destination host.
Subnet Masks
TCP/IP does not specify which portion of the IP address refers to the host and which portion to refer to the network. A router can determine whether the destination host is a part of the local subnet using an additional 32-bit number, though. The subnet mask is this particular number. These are some examples of subnet masks:
255.255.255.0
So how do you use a subnet mask to divide an IP address into a network address and a host address?
We must take into account the binary representations of the IP address and subnet mask in order to respond to that query.
11000000.10101000.01111011.10000100 (192.168.123.32) 11111111.11111111.11111111.00000000(255.255.255.0)
Now, it is possible to determine the network portion of the IP address by simply looking at the locations where the subnet mask is set to 1. Similarly, the IP address places where the subnet mask is set to 0 define the host section.
In this illustration, the last 8 bits of the IP address (10000100 = 32) refer to the host, whereas the initial 24 bits(11000000.10101000.01111011 = 192.168.123) refer to the network.
Routing
As a data package flows through a network (e.g. the internet) it is forwarded to it’s A data package is forwarded to its intended location via routers as it travels over a network (such as the internet). By taking into account the network portion of the IP address, the router may determine which network the host is a part of. To decide how to forward the packet to its destination network the best method possible, it consults its routing table. When the local router reaches the destination network, it considers the host portion of the IP address and sends the packet to the host.
IPv4 and IPv6
The words IPv4 and IPv6 are frequently used in network communication research and development. Despite the nomenclature, they refer to the first and second generations of the IP rather than the fourth and sixth: The fourth version of TCP is used for the first time by IPv4, the official Internet Protocol version. The immediate replacement for IPv4 is IPv6. Since the project was abandoned, IPv5 was omitted. The following version was given the name IPv6 to avoid misunderstanding.
Different address systems are used by IPv4 and IPv6 to identify devices in a network. While IPv6 may specify up to a sextillion (1036 = 1 with 37 zeros) addresses, IPv4 can only support up to 4 billion (4294967296 = 232) distinct addresses.
IPv4 Header
As was already explained, the Internet Protocol uses an IP header to represent the data required for effective routing. The IP header consists of 32 binary bytes. It can be divided into various fields that each describes a particular type of data.
Simple Usage (C++)
Below, you can see a straightforward example of a client-server configuration that makes use of TCP/IP. A straightforward TCP/IP connection between a server and client will be established by the code. The server that will display and respond to the message after receiving it will accept messages from the client. At the bottom of the page is a link to the complete script. This code is based on Sloan Kelly's excellent tutorial.
Create A Socket
The "receiving endpoints" of a link within a network are sockets. Client-server topologies are frequently employed, in which a server socket interacts with one or more client sockets.
// main.cpp
Std::cout <<"Creating server socket..." <<std::endl;
int list = socket(AF_INET, SOCK_STREAM, 0);
if (list == -1)
{
Std::cerr << "Can't create a socket!";
return -1;
}
The constants AF INET and SOCK STREAM specify the type of connection that must be made. The IPv4 address family is represented by AF INET, and a TCP connection is represented by SOCK STREAM.
Bind The Socket
The socket has to be instructed to connect to a certain address and port. The socket is "bound" throughout this operation. The socket address object hint is being created as a result.
// main.cpp
struct sockaddr_in hint;
No, we must appropriately configure the socket address object. We must specify the IP address, port, and the appropriate IP address family (AF INET).
// main.cpp
hint.sin_family = AF_INET.
hint.sin_port = htons (54000);
inet_pton (AF_INET, "0.0.0.0", &hint.sin_addr);
Here, the IP address "0.0.0.0" is used, which isn't technically a specification but instead lets the server choose the address it wants to use to accept inbound connections
The human readable port is converted into a format that the network can understand using the function htons, which converts an unsigned integer into a network byte order. A function called inet pton converts an IP address from text to binary.
// main.cpp
std: cout<< "Binding socket to sockaddr..." <<std: endl;
if (bind (listening, (struct sockaddr *) &hint, sizeof(hint)) == -1)
{
std:cerr << "Can't bind to IP/port";
return -2.
}
Let the Socket Listen
We must instruct our socket to listen for incoming connections now that it has been set up, connected to an IP address, and assigned a port:
std: cout<< "Mark the socket for listening..." <<std:endl;
if (list (list, SOMAXCONN) == -1)
{
std: cerr<< "Can't listen!";
return -3;
}
The socket's maximum allowable number of incoming connections is specified by SOMAXCONN.
Create a Client Socket and Start a Call
We're going to build a client that will acknowledge the connection from the server:
// main.cpp
sockaddr_in client.
socklen_t clientSize = sizeof(client);
std: cout<< "Accept client call..." << std::endl;
int clientSocket = accept (listening, (struct sockaddr *) &client, &clientSize);
std: cout<< "Received call..." <<std:endl;
if (clientSocket == -1)
{
std: cerr<< "Problem with client connecting!";
return -4;
}
Here, accept is being used to create a client socket. The listening socket that the client should connect to and an empty socket address object are both required by this function. It will give back a freshly made client socket that symbolizes the connection side of the client. It populates the IP family, address, and port that the client's connection is set to in the socket address object 'client'.
By using the function inetntoa, which transforms an IP address into a format that can be read by humans, we are able to keep track of the client's connection information.
// main.cpp
std:cout << "Client address: " << inet_ntoa(client.sin_addr) << " and port: " << client.sin_port << std::endl;
We can shut the listening socket now that the server and client are successfully connected:
Exchange Messages:
Let's converse! We'll build a reading buffer and wait for the client to send a message before we really exchange messages. Once a message has been sent by the client, the server stores it in the buffer before printing it and sending it back to the client.
The end of that. Now, utilizing a TCP/IP connection, we can transmit messages from the client to the server and back.
Full Script / Usage
Below is the complete script. Utilize it by compiling and running it with your preferred C++ compiler.
Open a terminal after the script has finished running and type:
telnet localhost 54000.
By doing so, a connection will be established between the client (your console) and the server script. To send a message to the server and receive it back, type it in and press enter.