Comments
episteme.arstechnica.com/eve/forums/a/tpc/f/469092836/m/138091881 - 95k
Posted by crouse
plz visit
www.itdojo.com/ccnaprep.htm - 15k -
Posted by waqasahmad
In the early implementations of ethernet, every device connected to a single wire. Thicknet (10-BASE 5) and Thinnet (10-BASE 2) were the most common physical layer implementations. A little later, hubs were used. All these technologies did effectively the same thing: connect many hosts together so that one of them at a time could transmit on the wire. This created a single, often large, collision domain. In these types of implementations, you can lose 50–60% of the available bandwidth just because of collisions. So if we had a 10-BASE T hub, not only did we actually end up with only about 4 or 5Mbs instead of 10Mbs, but that reduced bandwidth must also be shared by all the devices on that segment, instead of each device getting the full 10Mbs. Breaking up (segmenting) collision domains is necessary to make them small enough so that devices can reliably transmit data. We can segment using routers, but routers are expensive and difficult to configure; in addition, they don't typically have very many ports on them, so we would need a lot of them to segment effectively.
Bridges were developed to address this issue. A bridge isolates one collision domain from another while still connecting them and selectively allowing frames to pass from one to the other. A switch is simply a bigger, faster bridge. Every port on a switch or bridge is its own collision domain. The terms bridge and switch can be used interchangeably when discussing their basic operations; we use the term switch because switches are more modern and more common.
A switch must do three things:
• Address Learning
• Frame Forwarding
• Layer 2 Loop Removal
Posted by griffinLincoln
Please visit:
http://www.jlsnet.co.uk/index.php?page=ccna_1a_switching
Posted by sagitraz
-