Topology of physical connections

Causes

In the process of its evolution from batch processing systems built on the mainframe, computer networks became increasingly more complex, which gave rise to a large number of options for connecting and locating network components in space. This dictated the need to choose what the topology would look like when connecting more than two computers to the system.

Essence of topology

Topology — This is a graph whose vertices are end devices, and the edges act as communication lines.

A graph is one of the forms of representing a binary relation

The number of possible connections increases rapidly as new endpoints are added. The number of connections will mean topologically different connections. We also consider only graphs in which all vertices are united into a single network.

Three computers can be connected in only two ways (Figure 1).

Figure 1 - all possible connections of three terminal devices — Figure 1 *— all kinds of connections between three terminal devices*

The number of combinations with four terminal devices is six (Figure 2).

Figure 2 - various connections of four terminal devices — **Figure 2 – various connections of four terminal devices**

The number of combinations with five terminal devices is 21 (Figure 3).

Figure 3 - various connections of five terminal devices — **Figure 3 – various connections of five terminal devices**

End devices can be connected and combined as circumstances require, but in any case, transit nodes must have special means that transfer information from the sending node to the receiving node.

Before starting to design a network topology, you should evaluate the location and establish the tasks that will be solved. The solutions that will be used depend on this. For example, you can lay lines reserved for certain devices, create several paths between two nodes, which will distribute the load in case of active use and increase reliability, etc. A special place in network construction is occupied by the economic issue, which sometimes leads to the choice of simpler solutions and the minimum total length of communication lines.

Every modern computer network must meet four basic characteristics:

Scalability
Provides quick and easy network expansion if necessary.
Safety
Restrict physical access to network devices (for example, place them in a cabinet with a lock).
Prevent unauthorized access to devices (for example, put passwords on computers, install a firewall, restrict the connection of new devices, use intrusion prevention systems)
Secure information transmitted over the network (for example, use encryption, multi-factor authentication, password manager, access control lists).
Fault tolerance
Reduce the impact of failures by limiting the number of devices affected (for example, create multiple paths from the endpoint).
Quality of service
Using QoS. The topic of QoS is complex and vast, so it will be discussed briefly in order to explain the reasons for being at this point. This technology creates traffic priorities. Tagged QoS traffic is transmitted without queueing, while regular traffic is buffered.

Physical and logical topology

Physical topology is a topology that shows how nodes in a computer network are connected.

Logical topology is a topology showing how information will be transmitted on a computer network.

Usually, when they talk about topology, they mean the physical one, which will be discussed in this article, but do not forget about the logical topology, which is no less important when designing computer networks.

Fully connected and partially connected topology

A mesh topology is a topology in which every computer is connected to every other computer.

A loose-connected topology is a topology in which each computer is not connected to all the others and data exchange may require backhaul.

Fully connected topology is used very rarely, because it has obvious disadvantages (Figure 4). Each computer must have a large number of communication ports, the number of which will increase according to the formula where N — quantity end devices on the network. For each pair of end devices, a separate line must be allocated if it operates in full-duplex mode, and if it operates in half-duplex mode, then two lines are required, which makes this network construction option even more expensive. Another significant problem is the quadratic dependence of duplex lines on the number of nodes, the ratio of which can be represented as the formula where N is the number of nodes. The network has low scalability, which is dictated by the need to connect a new node to all network participants. Despite a decent number of shortcomings, this technology has its advantages, namely logical simplicity, high quality of service, since there is a direct channel for each node in the network, and high fault tolerance and quality of service, which is achieved due to a large number of connections.

Fully connected topology is used in multi-machine systems and networks with a small number of computers. The use of such technology in multi-machine systems is associated with high requirements for reliability and quality of service.

Figure 4 - Fully connected topology — **Figure 4 – Fully connected topology**

A partially connected topology allows you to create a large number of options for connecting network participants to each other, allowing you to choose the best option for a given case.

The main difference between a partly connected and a fully connected topology is that in the first, transit data transfer may be required to exchange data between two computers, but in the second, it is not.

Types of partially connected topologies

Communication redundancy is an increase in the number of communication lines to ensure increased fault tolerance.

Mesh topology is formed from a fully connected one by removing some connections (Figure 5). Despite the reduction in the number of connections, the disadvantages and advantages of this topology remain identical to the fully connected topology. The only thing that makes this technology stand out from the full mesh technology is the ability to connect a large number of end nodes, which is dictated by the absence of a quadratic dependence of the number of duplex lines on the number of computers. It is used in large networks or in combination with other partially connected topologies.

Figure 5 - Mesh topology — **Figure 5** **— Mesh topology**

Ring topology speaks for itself with its name. In it, computers are connected in series, but the last computer connected by this method is connected to the first, forming a ring (Figure 6).

Figure 6 - Ring topology — **Figure 6** **— Ring topology**

The ring topology is the most interesting because it has redundant connections, since each network participant can send data both clockwise and counterclockwise. Cyclicity allows you to check the delivery of information to the recipient. Leaving one computer, data can go full circle, notifying it of its delivery. This property of the topology can be used to find the network gap where the failure occurred. Two connections for a network participant increase fault tolerance, but if the node itself fails, the computer network will stop working. Network scalability is possible, but with an increase in network nodes, its fault tolerance will decrease and data transfer time will increase, so if the number of computers in the network increases by two, information will be forced to pass through another node. High quality service is provided by two ways for data transmission

This topology is used in fiber optic networks and Internet providers to organize reverse data transmission connecting central offices or headends.

Star topology consists of computers connected directly to a common central device. This approach allows for the centralization of services, which minimizes management costs and excess resources (Figure 7).

Figure 7 - Star topology — **Figure 7** **— Star topology**

This approach to creating a network provides quick and easy scalability, because to connect to a computer network you only need to draw another line from the central device, which is a switch (in older networks there could be a hub). In such a network, the following solutions have been thought out to increase its fault tolerance. Due to the fact that all devices are connected to a central point, all breakdowns such as cable breaks and equipment malfunctions are localized. Also, in such a network it is easy to detect a faulty segment and restore it without causing disruptions in the network, which follows from the first solution to increase fault tolerance. For such a network, which includes a large number of nodes, it is typical to use backup connections. The star topology reduces network congestion and reduces data transfer time due to the direct interaction of the device with the central element. The quality of service compared to the ring topology is lower because there is only one cable for data transmission and the central element is a bottleneck for data transmission, since if all network participants simultaneously start transmitting their data, it will either be discarded or end up in a buffer , therefore, when organizing such a topology, it is worth paying special attention to the parameters of the buffer memory of the network device so that it is enough to create a queue in case of overloads.

Star topology is used in high-speed local area networks because of its ease of maintenance, when creating multi-level networks, each level of which is a star, in networks that use twisted pair cables to connect devices.

Topology hierarchical star (extended star) or tree — the most common solution for creating computer networks in our time, both in local networks (LAN) and wide area networks (WAN). The topology is formed from an incompletely connected network, a star, by connecting new nodes that do not have direct contact with the central element (Figure 8).

Figure 8 — Hierarchical star or tree topology

This approach to network design has disadvantages and advantages almost similar to those of a star. In such networks, usually only network devices to which end users connect directly interact with the central element.

Hierarchical star is used in small, large, huge enterprises and offices. An example of using a hierarchical star for a plant consisting of two buildings, a workshop and an office (Figure 9).

**Figure 9 — Hierarchical star topology**

Common bus topology is a special case of star topology (Figure 10).

Figure 10 - Bus topology — **Figure 10** **— Bus topology**

Instead of a central element, a passive cable is used to which the end devices are connected. The wire becomes a common medium, that is, accessible to everyone, as does the information that spreads along it. Such a network has good scalability, since to connect a new computer you only need to connect it to a common cable; sometimes there is a need to extend it, which is easy to implement. The topology stands out from the rest because it is cheap, but it has significant disadvantages: any breakdown of the common cable paralyzes the network, which indicates low fault tolerance, low data transfer speed due to the fact that only one computer can use the common medium for transmission at a time (this ensures high security level, but also limits the number of network participants). For the network to operate, two terminals must be installed at the ends of a common cable. Another network problem is high packet loss. Due to these disadvantages, this topology has a low quality of service.

The bus topology can be used as a segment of a large network in which all computers are connected to a coaxial cable. Many wireless networks have a bus topology, with the radio medium acting as a common cable.

Mixed topology

A network segment is a logically or physically separate part of a network.

In small networks, star, bus, and ring topologies prevail, while large networks usually combine several topologies (Figure 11). For example, a hierarchical star topology may contain segments such as ring, star, bus, and others.