I'm studying for my CCENT exam, and not quite sure why the book (official Cisco) says that there are 5 collision domains in this example. Isn't it supposed to be 4? The hub on the left, the hub in the middle, between switch and router, and the switch on the right. And what exactly does the author mean by "Example of a Hub Not Creating Multiple Collision Domains, While Others Do"?
3 Answers
not quite sure why the book (official Cisco) says that there are 5 collision domains in this example. Isn't it supposed to be 4?
No, there are five collision domains. each bridge/switch interface is a separate collision domain, as is each router interface. The bridge in the drawing has two collision domains, and the switch has three; that gives you five collision domains.
And what exactly does the author mean by "Example of a Hub Not Creating Multiple Collision Domains, While Others Do"?
A hub is a layer-1 device, basically a powered cable, and all the hub interfaces are in the same collision domain, so a hub does not create multiple collision domains, while bridges, switches, and routers do.
The advantage of bridges and switches over hubs is that they reduce the number of collisions. They do that by creating more collision domains.
-
2I haven't seen a hub anywhere since 1999, and that was a hand-me-down from the university to the student union. May 6, 2019 at 3:28
A collision domain spans the nodes connected by a shared medium (=ports in half-duplex mode), a repeater (hub) or a combination of those.
A half-duplex port on a switch also belongs to a collision domain but it ends there (due to the switch's buffering). When full-duplex ports are used there is no collision domain. Full duplex isn't possible with inherently half-duplex (obsolete) media like coax with 10BASE5 or 10BASE2.
In your diagram the collision domains are separated by the bridge, the router and the switch(!). Assuming half-duplex operation for PCs 3 and 4, each has its own collision domain with its switch port.
If all ports are using full-duplex mode where possible, there are only two collision domains (those around the hubs). With full-duplex links there's only one transmitter and one receiver for each direction, making collisions impossible. The same goes for the router-switch link.)
Note that the scenario in that diagram is historical only. Neither repeater hubs nor half-duplex connections are used any more in professionally managed networks.
-
Based on the diagram, can I assume that PC3 and PC4 on different VLANs? May 5, 2019 at 14:40
-
3@AirWizardo, no, you do not have enough information to make that assumption. It doesn't matter if they are in the same or different VLANs, they are connected to a switch, so they are in different collision domains. The VLAN question would be about if they are in different broadcast domains. Each VLAN is a broadcast domain.– Ron Maupin ♦May 5, 2019 at 14:43
-
1@AirWizardo There's no indication for VLANs nor has that anything to do with collision domains.– Zac67 ♦May 5, 2019 at 14:43
-
Thank you both for the explanation, greatly appreciated. Everything makes sense now. May 5, 2019 at 18:04
@Air Wizardo the hub in put in there to intentionally deceive you. No body in the right mind would do that (unless they could not find a long ethernet cable and their only option was a hub unlikely)
And what exactly does the author mean by "Example of a Hub Not Creating Multiple Collision Domains, While Others Do"?
I am assuming this is Wendell Odom. In any case they mention Hub as a repeater with no intelligence. This was an old illustration and it perfectly displays a collision domain between Hub vs Switch. That is why Hub definitely needs CSMA/CD. If you understand this diagram and what it represents the concept of collision and broadcast domain would be smooth sailing for you. If you think carefully the "wifi-component" of your wireless router is a hub, and since there is no loop back mechanism over the air for detection of collision, WiFi does CSMA/CA (Collision avoidance)