You likely come across the term “collision domain” often when discussing different networking techniques. What is this phenomenon? What differentiates this domain from a broadcast domain? We answer your questions.
What is a collision domain?
A collision domain (collision domain in English) is a closed network composed, among others, of Layer 1 lines and stations in the OSI model.
A collision domain is characterized by the ability of a single device to send data over a physical transmission medium. As soon as at least two devices attempt to transmit data, a collision occurs, and the data to be transmitted therefore does not reach its destination. Indeed, if several devices attempt to access the support, the tension exerted on it increases and the different signals overlap. The more devices accessing the transmission medium simultaneously, the greater the chance that a collision will occur. If the distance between the different stations is too great, this can also encourage collisions.
Detecting and avoiding collisions: how does the CSMA/CD method work?
Carrier Sense Multiple Access/Collision Detection is a proven method for detecting and avoiding collisions. The acronym CSMA/CD is frequently used to refer to it. In French, this term means “Multiple access with carrier monitoring/Collision detection”. As for the Ethernet protocol, this flow control is governed by the IEEE 802.3 standard.
With this method, stations integrated into the domain control the voltage applied to the support. From the moment two stations send data simultaneously, the signals overlap and cause the previously defined limit values to be exceeded. This corresponds to the correct detection of the collision, and the sending process therefore repeats after a short latency. However, if the medium is free, then data signals can be sent without interference.
What are the boundary conditions for the size of the collision domain?
To ensure the correct functioning of the CSMA/CD method, the corresponding Ethernet protocol must not take into account more than 1,023 stations connected to the domain concerned. It is also not possible to use more than five cable segments and four repeaters, and only three segments can be connected to terminal stations. If these maximum values are exceeded, the collision domain is then too large, which no longer allows collisions to be detected using the procedure presented above.
How to determine a collision domain?
To delimit a collision domain, it is necessary to take into account the network segments that are part of it. Here, bridges and switches generally act as separation stations. These two components are coupling elements located at level 2 of the OSI layer. They support the transmission of data packets. Typically, switches can access more than two ports. Each of these ports is connected to a station by a cable; it then forms its own collision domain with it.
The situation is different when a hub is part of the network. Unlike bridges and switches, hubs operate on OSI layer 1. They connect several network stations together and their function can be compared to that of a distributor. The hub and all stations associated with it form a single collision domain. In such a constellation, only one station can exchange data signals with the hub each time. If another device tries to transmit data to the hub, it does not reach its target.
Disadvantages of collision domains
Outright collisions are practically inevitable on a daily basis on shared media. At first glance, they do not constitute a serious problem. To do this, however, they must not occur at too high a frequency. A high collision rate can indeed disrupt data traffic, requiring transmission to be interrupted. For this reason, the collision domains are not very scalable, if at all, because too frequent collisions cause a network overloadwhich causes considerable performance losses.
This problem is a disadvantage because economies of scale are essential to the long-term success of a business (the key word here is “economy of scale”). In principle, the CSMA/CD method is an effective solution for detecting collisions, but protocol management and maintenance can sometimes be quite complex.
What are the differences between a collision domain and a diffusion domain?
To begin with, collision domains and diffusion domains are two phenomena that relate to the logical division of a network. They are, however, characterized by differences in their level and their delimitation.
A broadcast domain is a cluster that encompasses one or more collision domains. A broadcast domain sends and receives some broadcast data traffic. Two components can be used to delimit it: a VLAN or a router. When two terminals are connected by routers, bridges or switches, there cannot be a collision between them; a collision can nevertheless occur between the terminal and the switch. Indeed, collision domains and diffusion domains operate on different layers of the OSI model.
Do you still lack experience in the field of network technology and would you like to learn some basics? If this is the case, we strongly recommend the following articles:
