Basics of networking

Let us now discuss the evolution of Ethernet. The initial versions of the standard, called thick networks (so-called thick Ethernet) and thin networks (so-called thin Ethernet), had few capabilities compared to what we have today. The older versions operate on a copper transmission medium (coaxial cable). They use a physical bus topology, which is characterised by all devices being connected to a common medium. The solution requires media access control, which is implemented using the CSMA/CD approach

After many years of using solutions based on bus topology as a transmission medium, it turns out that this solution is no longer efficient enough. The rapid growth of the network has led to ever higher user demands for bandwidth and reliability. Instead of coaxial cables, twisted-pair cables, UTP cables and new topologies are widely used. Star topologies appeared, the same ones used today, but using hubs instead of switches as the central point of the network. No one had heard of switches then.

The use of hubs improved the performance of computer networks to some extent, but it soon became apparent that this solution was not ideal either. The basic feature of a hub is that it transmits data to all the devices connected to it. It works like this, that a computer that wants to send data to another device performs this communication through the hub. The latter, on the other hand, is not so clever as to transfer data to the appropriate device, it simply sends data to all those connected to it.

Only the devices to which the data is sent analyse the addressing to determine whether they are recipients. If they are not recipients, they ignore the data, and if they are, they interpret it.

This type of solution means that although the physical topology is a star topology, it is logically similar to that used in the previous generation of Ethernet. Here, too, a link access method based on CSMA/CD is used, which has become inefficient due to the rapid growth of the network. In addition, each hub creates a so-called collision domain.

The more devices connected to the hub, the larger the collision domain, and the larger the collision domain, the greater the likelihood of collisions, limiting throughput and creating requirements for frequent data retransmissions. More collisions are not the only problem associated with using hubs. Other disadvantages of such devices include limited scalability and increased delays in data transmission, among other things due to the aforementioned shocks.

Efforts to address the weaknesses of hub-based Ethernet continued over the years until the invention of a smart networking device called a switch, which solved the problems that plagued earlier versions of Ethernet.

Switches in computer networks are still around today and there is no indication that this will change any time soon. Why are these devices so popular and why are they so smart? Well, unlike a hub, a switch does not send data to all the devices connected to it, but only to the specific device for which the data is destined, obviously bypassing broadcasting, such as the ARP transmission discussed earlier. A logical point-to-point topology exists between the switch port to which the device is connected and the device itself. Data sent to a particular device is sent to it and only to it.

The use of a switch almost completely eliminates the risk of collisions, as devices do not have to compete with each other for access to the medium. At the same time, the size of the collision domain is limited, as such a domain consists only of the switch ports and the devices connected to it. There are many more advantages of switches. Each device connected to a switch port has a dedicated bandwidth available. For example, if a switch offers a transfer rate of 100 Mbps, this bandwidth will be available to each device connected to it.

With a hub, this bandwidth is shared between all devices. By using a switch, data can also be transmitted in full-duplex mode, which means that the devices connected to it can receive and send data simultaneously.

There are several versions of the Ethernet standard in use today. The most popular of these is the standard offering nominal throughputs of up to 100 Mbps, known as the FastEthernet standard. Transmission in this standard is over only 2 copper pairs rather than 4 twisted pairs. It is a common solution used in many computer networks. In most cases it meets the requirements of computer networks.

The Gigabit Ethernet standard can be used when the demand for network bandwidth increases with the amount of data being transmitted. Nominally, it provides a throughput of
1 Gbps. If the 1000BASE-T standard is used, all copper twisted pair cables are used for transmission. This version of Ethernet is used by large local networks that use VoIP telephony and transmit large amounts of various types of media.

Using the Ethernet standard, data can also be transmitted over fibre optic links, in which case the Gigabit Ethernet standard is called 1000BASE-SX or LX. There are also Ethernet standards that provide communication at 10 or even 100 Gbps. They are mainly used in metropolitan and wide area networks because they are very, very expensive to implement and few people can afford to use this type of solution in a local area network. The table below shows the most popular versions of the Ethernet standards and the transmission medium they use:

Ethernet standard	Maximum throughput	Transmission medium used	Maximum distance
100BASE-TX (fastEthernet)	100 Mbps	UTP (cat. 5/5e)	100 metres
100BASE-FX (fastEthernet)	100 Mbps	Optical fibre (single/multi-mode)	400/2000 metres
100BASE-T (gigabitEthernet)	1 Gbps	UTP (cat. 5e)	100 metres
100BASE-TX (gigabitEthernet)	1 Gbps	UTP (cat. 6)	100 metres
100BASE-SX (gigabitEthernet)	1 Gbps	Multimode optical fibre	550 metres
100BASE-LX (gigabitEthernet)	1 Gbps	Single-mode optical fibre	2000 metres
10GBASE-T (10gigabitEthernet)	10 Gbps	UTP (cat. 6/7)	100 metres
10GBASE-LX4 (10gigabitEthernet)	10 Gbps	Single-mode/multi-mode optical fibre	300/10000 metres

 

The switches described above use MAC addresses to transfer data between devices connected to the switch ports. Every switch has something called a MAC address table. This is nothing more than a collection of information that determines which device, actually what MAC address of a device is connected to a particular port. 

Entries in such a table are added dynamically and not by the administrator. The switch retrieves the information stored in the table during the learning process. From a received frame, the switch reads the source MAC address and adds it to its table, assigning the port number on which it received the frame. In turn, if it does not know to whom to send such a frame because there is no entry for the recipient's MAC address in the table, a process called flooding occurs.

This can be compared to broadcasting, as the frame is sent to all devices except the sender. The device to which the frame is not addressed discards it, while the receiving device responds and sends the frame to the switch. The switch reads the MAC address of the sender from the frame and stores it in its table. The whole process of learning and flooding is shown in the video tutorial.

Ethernet frame

Since the Ethernet standard operates on the second layer of the OSI model, you can guess,
that it also creates its own frames. Of course, yes, Ethernet encapsulates its own frame, called an Ethernet frame. You can see an example frame below:

Field size in bytes	7	1	6	6	2	46 - 1500	4
Field name	Preamble	Frame start marker	Recipient's MAC address	Sender MAC address	Length/Type	Data and filling	Frame Control Code (FCS)

 

·         Preamble and Frame Start Marker - these fields are used to inform the target device that it is ready to receive frames;

·         The target MAC address, which is the physical address of the recipient of the frame;

·         The source MAC address, which is the physical address of the sending host;

·         Length/Type - The length field specifies the size of the frame, while the type specifies the protocol used by the higher layers, the most common of which is IPv4;

·         Data - this is the packet received from the network layer. The minimum size of this field must be 46 bytes and the maximum size must be 1500 bytes. If the packet is smaller than 46 bytes, it is supplemented with random data to increase the size of the entire frame to the required minimum, i.e. a maximum of 64 bytes.

·         Frame check code - field containing frame checksum, used to detect possible frame errors. The device sending the data calculates the checksum and places it in the frame, the data receiver also calculates the checksum after receiving the data; if both checksums are correct, the frame is accepted, if they are different, the frame is considered damaged and rejected.

The total frame size can be up to 1518 bytes (the preamble and start of the frame signal are not taken into account when calculating the frame size). There is also an Ethernet frame
frame with a maximum length of 1522 bytes. Such frames are used in virtual LANs,
in so-called VLANs.