CCENT - Full vs Half Duplex

Daniel333Daniel333 Member Posts: 2,077 ■■■■■■□□□□
Hey guys,

Been about 3 years since I have done any Cisco. And what I have done wasn't at the LAn level. All my cert expired and i need to do them all from scratch.

It seems to me, when two devices are on a Hub we can still have full duplex. Until a third device enters the mix, the NICs should auto-detect themselves as full deplex right? There is no collision to detect?

So I am thinking ONLY the second example would go into half-duplex mode?

-Daniel

Comments

  • EdTheLadEdTheLad Member Posts: 2,111 ■■■■□□□□□□
    If all nic's support full duplex, they will become full duplex, due to the hub, they will have collisions and you will see dropped frames from collisions etc.
    Networking, sometimes i love it, mostly i hate it.Its all about the $$$$
  • Daniel333Daniel333 Member Posts: 2,077 ■■■■■■□□□□
    Thanks for your reply. I really appreciate it. Going to test this on physical hardware tonight. Been at work so I am just sorta playing on a laptop here on breaks. Amazing what you can forget in a few years.

    Feeling silly here Googlein' around, but I am trying understand the details on CDMA here, which bit is that causes the collision, which causes the auto-detect switch to half.

    Example 1 - From the image above - Until I add a third item to a hub, there should be no collisions right? Therefore making the HUB a repeater functioning at 10/full duplex.
    Example 2 - SHOULD get a collision eventually and autodetect should throw the NICs into half.
    Example 3 - there are again, only two devices. So I should be seeing full duplex, 10meg.

    I guess the only way I could see the NIC realizing its on a hub is if anything it sends out it's own send plane, cmes back on its own receive plane. But I don't believe that happens in a HUB or repeater.

    That sound about right?
    -Daniel
  • [Deleted User][Deleted User] Senior Member Posts: 0 ■■□□□□□□□□
    The hub would make everything half duplex that is connected to it. All devices connected to a hub would communicate in half duplex. The devices connected to the switch should communicate in full duplex because switches support CSMA/CD. You could set the port to run in either full or half duplex by using the duplex command in interface config mode I believe. Just run a show interface ___ to check the duplex also.
  • EdTheLadEdTheLad Member Posts: 2,111 ■■■■□□□□□□
    The hub would make everything half duplex that is connected to it. .

    Really? How do the devices know they are connected via a hub rather than a cable?
    Networking, sometimes i love it, mostly i hate it.Its all about the $$$$
  • [Deleted User][Deleted User] Senior Member Posts: 0 ■■□□□□□□□□
    Well a cable is implied. I don't think that Cisco makes a wireless hub. I may be wrong.
  • EdTheLadEdTheLad Member Posts: 2,111 ■■■■□□□□□□
    Since you don't know the options available, you could connect two devices directly via a cable or use a hub as a physical layer repeater in between obviously using cables, but for some maybe it's not so obvious. The end devices send pulses to negotiate duplex, since these pulses are just relayed by the hub, both sides can be full-duplex.
    Networking, sometimes i love it, mostly i hate it.Its all about the $$$$
  • Daniel333Daniel333 Member Posts: 2,077 ■■■■■■□□□□
    Thanks again guys.

    That is exactly what I am trying to understand. The internal workings of how it determines its on a hub. vs a repeater vs straight cable.
    -Daniel
  • Daniel333Daniel333 Member Posts: 2,077 ■■■■■■□□□□
    Alright, so setup some hardware and learned a few interesting things.

    MacMini 2006 > Netgear Hub < MBP2008

    The INSTANT I plug it in the netwok interface page switches to half-duplex at 10meg. So my original theory was detection was based on what the NIC read of the hubs receive plane. But after reading this

    Autonegotiation - Wikipedia, the free encyclopedia

    I realized I am mistaking autonegotiation with CSMA/CD. Here is the operative line from the wiki " These connection present pulses are sent by Ethernet devices when they are not sending or receiving any frames. They are unipolar positive-only electrical pulses of a nominal duration of 100 ns, with a maximum pulse width of 200 ns,[3] generated at a 16 ms time interval (with a timing variation tolerance of 8 ms)."

    So! Yes, instant you plug something ethernet in its receives these unipolar pulses which determine what mode to be in.

    SOO! Does this mean we CAN'T set the ethernet settings to full duplex since we KNOW there can be no collisions. So I went into both macs and set 10meg full duplex. So far so good.

    Suppose level 2 is going to be to transfer file in both directions at the same time and monitor for lost packets and measure transfer time? Thoughts?
    -Daniel
Sign In or Register to comment.