Collision domain

from Wikipedia, the free encyclopedia

The term collision domain denotes a sub-area of ​​subscriber stations in the same OSI model layer 1 in a computer network. A collision domain comprises all network devices that compete for access to a common transmission medium. The transmission medium is therefore a resource shared between all network stations. The basic idea is that all network participants have the opportunity to use the network equally.

With a common medium, only one station can transmit information at a time, which is transmitted to or received by all other stations. Catch in such a common layer-1 - segment two stations at the same time to send, there will be collisions . They arise because both stations work on the same physical medium (cable or radio frequency). The signals (voltage pulses) are mixed / superimposed in the medium and the information is thereby destroyed.

A collision domain is a network segment in a CSMA / CD network. All stations that are connected to one another on a common layer 1 (physical layer, bit transmission layer) either directly or with repeaters or repeater hubs are located in a common collision domain.

Bridges separate collision domains because they work on Layer 2 and cache packets (frames). In a switched network in half-duplex operation, a collision domain only consists of two ports, the client port and the switch port; with full duplex connections there is no longer a collision domain.

Formation of collisions

In a network built with coaxial cables (for example 10BASE2 and 10BASE5 ), often all stations are actually connected to a single cable. If a repeater is added, the common collision domain still remains, since repeaters only process signals electrically and do not react to collisions. A 10BASE-T - Ethernet looks like a star, but when hubs are used - from a logical point of view - it still consists of a single medium to which all stations are connected. All stations are in a collision domain. If at least two stations start transmitting at the same time, a collision occurs. With Ethernet, attempts are made to avoid collisions and minimize their effects with the Carrier Sense Multiple Access / Collision Detection process (CSMA / CD).

PC1 <----------> PC2 <----------> PC3 <----------> PC4

Such a network can be divided into different segments and thus into collision domains with a bridge , for example .

----------> PC1 <----------> PC2
                                     PC3 <----------> PC4

PC1 and PC2 as well as PC3 and PC4 and the respective side of the bridge are each located in a separate collision domain. This means that PC1 can communicate with PC2 and PC3 can communicate with PC4 at the same time. Collisions can only occur within a segment: When PC1 sends something to PC3, the packet is first temporarily stored in the bridge and the destination address is read. Then the bridge sends the packet on the lower segment. If PC3 or PC4 send something at this moment, a collision occurs in the lower segment. Compensating for this collision and retransmitting the packet is now up to the bridge, not PC1.

The next step in the development is therefore to attach each PC to its own port on the bridge (microsegmentation). A bridge with more than two connections is usually referred to as a switch . If all devices involved are capable of full duplex , no more collisions occur:


Handling of send conflicts

There are two ways to deal with this problem:

The CSMA / CD process is becoming less important. Today's network architectures are based on microsegmentation, which only connects one station to a switch port and consequently no longer creates a competitive situation. The method is no longer included in the current 10 Gigabit Ethernet standard. In contrast, the concept of the CSMA / CA process is still an indispensable part of WLANs, since all stations use the same, common medium.