Why are computer network protocols required

Adviser: what is what with the network protocols

When talking about network protocols, the term OSI layer model or OSI reference model comes up quickly. OSI stands for Open Systems Interconnection, and the reference model is the design basis of communication protocols in networks. The model knows seven different layers. The recipient and the sender must work according to certain rules that are laid down in protocols. This is the only way they can agree on how data should be further processed.

The layers of the OSI model can be roughly divided into transport and application layers. The first four layers belong to the transport-oriented layers: Here you will find protocols such as IP, IPsec, IPX, SPX, TCP and UDP. Other protocols are at home in the, for example FTP, HTTP, HTTPS, SMTP. The following overview gives you a brief and concise overview of the most important network protocols.

Appletalk and Ethertalk

Appletalk is standard on Macintosh computers. It is not a single network protocol, but a group of them - for accessing files or printers, for example. If necessary, you can use Appletalk to establish a connection between two computers without an Ethernet cable: via the serial interface. The connection is then made via printer cables, which are also connected serially on Macintosh computers. Several computers can be connected via one device that loops through the network connection. If Mac computers are connected via Ethernet, one speaks of Ethertalk.

Appletalk was abandoned with the operating system version Mac OS X 10.6 in favor of IP-based networks. Finding devices and servers in the network is done by Bonjour, for example.


DHCP (Dynamic Host Configuration Protocol) simplifies the assignment of IP addresses in a network. At least one DHCP server is required for this. It distributes IP addresses to the various clients. An address space is specified on the server side. IP addresses from this pool are assigned to the clients as soon as they ask the DHCP server to do so. For this to happen, the network configuration of the clients must be set accordingly. This is the standard nowadays with operating systems such as Linux and Windows, so that an administrator does not have to worry about the client settings.

A DHCP server assigns IP addresses automatically, dynamically, or manually. Addresses can only be assigned once automatically. A client identified by the MAC address of its network interface receives a free IP address the first time it logs into the network. This is linked to the MAC address and the client will receive it every time he logs on in the future. In contrast, dynamically distributed IP addresses are collected from the server after a certain period of time.

The method should definitely be used in a larger network, otherwise the IP addresses could become scarce there. Caution: This can even happen in smaller networks if you often use virtual environments such as VMware Workstation, Parallels Desktop or Citrix XenServer. Depending on the configuration, these generate several virtual network interfaces. The most cumbersome method, but one that is definitely used, is the manual assignment of IP addresses. Here an administrator can assign a very specific address to one or more clients - for example a router or a network printer.

The request from the clients takes place via broadcast. This tests how many DHCP servers there are in the network. The person addressed sends a configuration proposal in response. If the client accepts this, it sends a DHCP request. In return, the server sends the IP address and the required parameters. At the end of the session, the client sends a release back to the server, reporting that the IP address is no longer required (DHCP release).


FTP (File Transfer Protocol) is the standard format for exchanging files over the Internet. By default, an FTP session takes place via port 21. The protocol is mostly used to access file archives. As a rule, users without an account can also access the Internet server (anonymous FTP). In other cases, authentication is carried out using a user name and password. FTP knows almost 80 commands. The server is accessed via graphically oriented programs, via an Internet browser or directly via a shell. To do this, the user simply enters "ftp IP ADDRESS" or "ftp HOST" on the console. "Ftp>" usually appears as an input prompt. You then get a list of all commands with "?" or "help". A short explanation of the individual commands can be found with "? COMMAND" or "help COMMAND". Important commands are:

CD: changes the directory

close: cuts the connection to the FTP server

del: deletes a file on the FTP server

to you: lists the directory contents

exit: ends the FTP session

get: Download file

ls: lists how to you Directories

mdel: deletes multiple files on the FTP server

mget: download multiple files

mput: sends multiple files to the FTP server

open: connects to an FTP server

put: sends a file to the FTP server

pwd: shows the current directory on the FTP server

quit: ended like exit the FTP session

user: sends user identification

Data transmitted via FTP is not encrypted. If you want to use a secure method with FTP, you have two options: You can use FTP over SSL (FTPS); here the FTP connection is encrypted. Or you can transfer data via Secure FTP (SFTP); then the FTP connection is tunneled via SSH. You can use these options individually, or you can switch to SSH as the transfer protocol. Because SFTP has been included as a subsystem there since version 2, and you save a server on the host.


HTTP (Hypertext Transfer Protocol) is the underlying protocol for the World Wide Web. It runs over port 80 and transmits data from a web server to the web browser and vice versa; As with FTP, TCP / IP is used as the transport protocol. HTML documents are transmitted, but also images, audio and video data. How and with what this additional data is represented is determined by the MIME types on the client. HTTP is available in versions 1.0 and 1.1. These differ in some functions. In version 1.0, for example, the connection is terminated as soon as the data has been transferred. In version 1.1 you can maintain the connection with the header entry keep-alive. Version 1.1 can also resume aborted transfers.

HTTPS is available for encrypted transmissions - communication takes place via port 443 as standard. The protocol developed by Netscape shifts a layer between HTTP and TCP. HTTPS is identical to HTTP; the encryption is carried out using TLS (Transport Layer Security), previously known as SSL (Secure Socket Layer). In HTTPS, a secure authentication of the server and client takes place first. A session key is then generated and exchanged, which is used to encrypt the transmitted data. If you want to use an HTTPS server, you need SSL routines. These are loaded as a module in the Apache web server, for example. You also need a digital certificate, which is usually issued by a certification authority. The costs for this range from around $ 40 to over $ 1,000 per year. Some organizations like Startcom and Cacert also issue certificates for free.

IP, IPv4 and IPv6

IP (Internet Protocol) is the basis of the Internet. This itself consists of backbone routers and other networks in universities and with providers. You can create additional subnets and assign them to customers in the form of IP addresses. The data between the routers is controlled via IP. The main task of the IP is to put together small packets from the information units and to deliver them to their destination via various routes - i.e. from one IP address to another. Once there, they are reassembled in the correct order by the IP protocol.

The IP addresses of version 4 (IPv4) are composed of four octets. The fourth version of the Internet Protocol was the first to be used worldwide. Since then, the address spaces have become scarce. The last two free address blocks were assigned in February of the last year. This means that no further IP addresses can be distributed from the IPv4 address pool with over four billion unique IP addresses.

The direct successor to IPv4 is version 6, or IPv6 for short. In this IP version, addresses have a length of 128 bits instead of the previous 32 bits. The current operating systems already master this form of addressing. This is not surprising, as IPv6 has been the standard since 1998. Little by little, this new 6-version will replace the old 4, as it allows significantly more addresses: instead of around 4.3 billion (2 to the power of 32), IPv6 can handle around 340 sextillion (2 to the power of 128). The smooth transition is also achieved by adding IPv6 to the existing IPv4.

In addition, IPv6 offers further advantages: The IPsec encryption method is integrated, and the addresses can be configured automatically in such a way that methods such as DHCP are superfluous (experts speak of stateless configuration). Network Address Translation (NAT) is also superfluous: With IPv6, users receive globally unique IP addresses so that every device, from server to NAS, from tablet to smartphone, can have its own IP address worldwide - an address translation for its own network is therefore superfluous.

IPv6 addresses are noted in hexadecimal. The number is divided into eight blocks of 16 bits each. The blocks are separated by colons, leading zeros and zero blocks can be omitted. To avoid confusion with port numbers, IPv6 addresses are placed in square brackets in the browser: http: // [1020: de2: 74a1 ::: 9088: 7890: abcd]: 631 /.


IPX (Internet Protocol Exchange) and SPX (Sequenced Packed Protocol Exchange) are used by NetBIOS for data transmission. Similar to IP, the protocols split the data into packets and combine them again at the destination. SPX is an extension to IPX that also checks for errors. Both require the Netware operating system from Novell.


NetBIOS (Network Basic Input / Output System), as the standard interface on the transport layer, controls access to devices available in the network such as printers, scanners and streamers. It was developed for IBM in 1983 and contains functions for name resolution, among other things. NetBIOS is implemented on network cards and forwards requests from the user into the network.

The transport protocol based on NetBIOS is called NetBEUI (NetBIOS Extended User Interface). It is a Microsoft extension of NetBIOS. NetBEUI was developed for DOS PCs and was the standard under Windows up to version 2000 and ME. The log is small and fast. Since it is based on MAC addresses, it works without IP addresses; Source and destination computers are identified by their host name, which can be up to 15 characters long. However, it cannot be routed, is no longer up-to-date in larger networks and has not been available in Microsoft products since Windows Vista.


SMB (Server Message Block) or LAN Manager is mainly a protocol for file and print services. It runs over NetBIOS, NetBIOS over TCP / IP, and directly on the TCP / IP protocol stack. SMB implements an NFS-like file system and is therefore independent of the server's file system. The Samba project, for example, makes use of this, a free implementation of the SMB protocol under Unix and Unix systems such as Linux. With Samba, Windows servers can be replaced by cheaper Linux systems in heterogeneous environments.

SMB was developed at IBM in 1983. Subsequently, companies such as Microsoft, SCO and the Samba project team improved the protocol.


SMTP (Simple Mail Transfer Protocol) is the protocol for sending e-mail. The protocol from the application layer can be found on port 25. Alternatively, e-mails can now also be sent via port 587; there, e-mails known to users of the system are received and forwarded. This is to prevent unwanted spam. SMPT is primarily used to send emails; protocols such as POP3 and IMAP are used for receiving mail. An SMTP server is either available in the local network or can be reached via external servers - with a provider on its own website. So-called Mail Transfer Agents (MTA) then take care of the further transmission of the mails on the Internet.

SMTP servers communicate with each other in clear text. Since SMTP is text-based, you can, for example, easily send a mail via Telnet. That is also the reason why e-mail addresses are not considered reliable. Because in Telnet, as in other programs, the name of the recipient is just as freely selectable as that of the sender.


TCP / IP (Transmission Control Protocol / Internet Protocol) is so named because TCP is in most cases based on the Internet protocol. TCP / IP was originally developed for the US defense system to connect computers in different systems with one another. This also explains some properties such as the fact that data loss is recognized and automatically corrected or that an overload of the network is prevented. While IP takes care of sending the packets, TCP ensures the reliable data flow between two points. It checks the integrity of the data with the help of a checksum in the packet header and ensures the order using so-called sequence numbers. Incoming packets are merged at the recipient, duplicates sent are discarded.


UDP (User Datagram Protocol) is a simple, connectionless protocol. In contrast to TCP, UDP transmits data that does not necessarily have to arrive. In other words: with UDP there is no guarantee that data will arrive - with TCP there is. The information in the header of the packets is limited to the bare minimum. This also means that no congestion control can take place. The only thing that can be checked with UDP is the correct checksum after receiving a message. (hal)