Domain Question

typesh

Hello everyone... Just wondering if someone can confirm whether I am correct.

Thinking about broadcast/collision domains:

Routers break up both broadcast domains and collision domains.

Switches break up collision domains by placing each node on its own collision domain (provided only one device is connected to that switchport). If Router1 has Switch1 connected out its FastEthernet interface, and Switch1 connects to Switch2, and Switch2 connects to Switch3, then everything outside of the Routers FastEthernet interface is one broadcast domain, with collision domains occurring on the individual switchports. So if 3 switches are all connected, we do not have 3 broadcast domains- we still have 1. The number of collision domains would depend on how many devices are connected to these switches plus the number of trunk connections.

Hubs create one collision domain and one broadcast domain. Even if many hubs are connected, it is still considered one large broadcast/collision domain.

Can anyone please confirm? Thank you!

wastedtime

Yes, although I find it easier to look at what layer the devices work at. Layer 1 devices do not separate collision or broadcast domains. Layer 2 devices separate collision domains but don't separate broadcast domains. Layer 3 and above separate both collision, and broadcast domains.

Bl8ckr0uter

typesh wrote: »

Hello everyone... Just wondering if someone can confirm whether I am correct.

Thinking about broadcast/collision domains:

Routers break up both broadcast domains and collision domains.

Switches break up collision domains by placing each node on its own collision domain (provided only one device is connected to that switchport). If Router1 has Switch1 connected out its FastEthernet interface, and Switch1 connects to Switch2, and Switch2 connects to Switch3, then everything outside of the Routers FastEthernet interface is one broadcast domain, with collision domains occurring on the individual switchports. So if 3 switches are all connected, we do not have 3 broadcast domains- we still have 1. The number of collision domains would depend on how many devices are connected to these switches plus the number of trunk connections.

Hubs create one collision domain and one broadcast domain. Even if many hubs are connected, it is still considered one large broadcast/collision domain.

Can anyone please confirm? Thank you!

This is correct, by default. Don't forget about Vlans, which break up collision domains and broadcast domains. Just thought I should throw that out there since you're posting in the CCNA forums and not the net plus forums.

Virtual LAN - Wikipedia, the free encyclopedia
http://www.cisco.com/univercd/cc/td/doc/product/software/ios113ed/113ed_cr/switch_c/xcvlan.htm

So by default:
Hubs: Layer 1 - 1 col/1 broad
Switches: Layer 2- X col/1 broad
Routers: Layer 3- X col/X broad

With Vlans:
Switches: Layer2/3- Y col/Y broad

X= The number of ports on that device. For example a 12 port switch would have 1 broadcast domain but 12 collision domains.

Y= The number of Vlans running on that device. For an example, a switch with 4 Vlans set up would have 4 broadcast domains and 4 collision domains.

typesh

Thanks a lot! I got it clear now

Just another question though, regarding VLANs.

I am working on ICND1 stuff, and this may be and ICND2 question... but how does a switch know what VLAN a frame is a part of?

If we had 2 switches connected to each other, on each others Fa0/1 port (this will be the trunk port between the two switches). Lets take all the remaining ports on SW1 and leave them in VLAN 1. Lets take all the remaining ports on SW2, except for Fa0/2 which has a PC with the IP of 192.168.1.2/24, and place them in VLAN 200. When a PC1 who has an IP of 192.168.1.1/24 and is connected to SW1 sends out a frame, it will travel over the trunk to SW2. How does SW2 know to send the frame out Fa0/2 and not all the other ports? There is no Ethernet Header/Trailer field that supports something to identify VLAN# and the switch does not read IP addresses (its a L2 switch).

Thank you.

Edit... Now that I have asked that... It makes me think of something else. What if the other switch has no ports in a specific VLAN. It will still receive all VLAN traffic over the trunk port, even though it has no ports that correspond to certain VLANs. This seems like a waste of bandwidth..

EdTheLad

typesh wrote: »

If we had 2 switches connected to each other, on each others Fa0/1 port (this will be the trunk port between the two switches). Lets take all the remaining ports on SW1 and leave them in VLAN 1. Lets take all the remaining ports on SW2, except for Fa0/2 which has a PC with the IP of 192.168.1.2/24, and place them in VLAN 200. When a PC1 who has an IP of 192.168.1.1/24 and is connected to SW1 sends out a frame, it will travel over the trunk to SW2. How does SW2 know to send the frame out Fa0/2 and not all the other ports? There is no Ethernet Header/Trailer field that supports something to identify VLAN# and the switch does not read IP addresses (its a L2 switch).

Thank you.

Edit... Now that I have asked that... It makes me think of something else. What if the other switch has no ports in a specific VLAN. It will still receive all VLAN traffic over the trunk port, even though it has no ports that correspond to certain VLANs. This seems like a waste of bandwidth..

I'm not going to spoon feed you the answer.Do some reading on vlan tagging "dot1q" & "ISL" this will inform you on how a vlan tag is sent in a frame.Then do some additional reading on "Cisco Native Vlan".Once you read up on these topics i'm sure you will be able to answer your previous question.

mikej412

typesh wrote: »

What if the other switch has no ports in a specific VLAN.

You can configure VLAN pruning

typesh

EdTheLad wrote: »

Do some reading on vlan tagging "dot1q" & "ISL"

Nice! That's what I was looking for. The 12 bit field in the VID portion of the 802.1q explains where the 4094 comes from (always wondered why the highest VLAN number was 4094 when doing configuration)

Thanks EdTheLad

mikej412 wrote: »

You can configure VLAN pruning

This is really cool actually. Almost makes me think of ...routing protocols.

Thanks as well Mikej.