|
Home > Archive > alt.certification.cisco > October 2003 > Failed CCNA - Question re: Full/Half Duplex
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Failed CCNA - Question re: Full/Half Duplex
|
|
| Old Garb 2003-10-03, 4:24 pm |
| Well, my first recognized cert test today was a bit of a disappointment =
as I only scored a 770. At least now I know which areas to concentrate =
on. I also got a little balled up early on the first router sim which =
cost me some time and started to run short toward the end (20 questions =
remaining and 8 minutes left...)
There are NO IPX questions at all on this 604-607 exam (I didn't expect =
any) but... I had at least 3 questions on half/full duplex Ethernet I'm =
unsure of. I know full duplex uses it's own pair to transmit and a =
separate pair to receive thus dramatically reducing collisions but does =
it eliminate collisions altogether? I thought, no.
Secondly, they asked if full duplex incorporated it's own private =
collision domain. I just had never thought of it termed as a "private" =
domain.
| |
|
| Full duplex has no collisions. It has one transmit pair and one receive
pair, so the incoming and the outgoing traffic don't collide. For full
duplex to operate, it has to be connected to a switched port because the
CSMA/CD mechanism is disabled in full duplex. So, theoretically, there can
only be two hosts communicating with each other (each using one transmit and
one receive pair). There's no mechanism for a third host to join the
conversation because the first two hosts don't expect their transmission to
be interrupted. I'd say that each host has its own "private" collision
domain because collisions happen when two hosts start transmitting on the
same pair at approximately the same time. Since either pair has only one
transmitting host, there's no other host in the collision domain.
Grey
"Old Garb" <OSS@attnet.net> wrote in message
news:Srkfb.30155$%h1.18190@sccrnsc02...
Well, my first recognized cert test today was a bit of a disappointment as I
only scored a 770. At least now I know which areas to concentrate on. I
also got a little balled up early on the first router sim which cost me some
time and started to run short toward the end (20 questions remaining and 8
minutes left...)
There are NO IPX questions at all on this 604-607 exam (I didn't expect any)
but... I had at least 3 questions on half/full duplex Ethernet I'm unsure
of. I know full duplex uses it's own pair to transmit and a separate pair
to receive thus dramatically reducing collisions but does it eliminate
collisions altogether? I thought, no.
Secondly, they asked if full duplex incorporated it's own private collision
domain. I just had never thought of it termed as a "private" domain.
| |
| Scooby 2003-10-03, 5:25 pm |
| Yea, what Grey said. Right on.
For more clarification... All hosts use 1 pair for sending and 1 pair for
receiving (not just in full duplex). But, where there are only two hosts
(you to another device), they can both talk and receive at the same time.
But, if there are multiple hosts connected through a hub, then each person
talking will require everyone else to listen. If someone else starts
talking during that time, then everyone will be required to listen to two
conversations at once and they doesn't work (at least not electronically).
You can talk and listen, you just can't listen to two people at the same
time. That is where collisions come in - they say "back off, someone is
already talking".
So, if you are in a shared environment, only one person can talk at a time.
Meaning that if someone is talking to me, I'm not going to start talking
since it will cause collisions. When you know that there are only two of
you, you can go full duplex and you can both talk at the same time.
"Grey" <bbb@ccc.com> wrote in message
news:19-dnRAgI9kqROCiU-KYvQ@comcast.com...
> Full duplex has no collisions. It has one transmit pair and one receive
> pair, so the incoming and the outgoing traffic don't collide. For full
> duplex to operate, it has to be connected to a switched port because the
> CSMA/CD mechanism is disabled in full duplex. So, theoretically, there can
> only be two hosts communicating with each other (each using one transmit
and
> one receive pair). There's no mechanism for a third host to join the
> conversation because the first two hosts don't expect their transmission
to
> be interrupted. I'd say that each host has its own "private" collision
> domain because collisions happen when two hosts start transmitting on the
> same pair at approximately the same time. Since either pair has only one
> transmitting host, there's no other host in the collision domain.
>
> Grey
>
> "Old Garb" <OSS@attnet.net> wrote in message
> news:Srkfb.30155$%h1.18190@sccrnsc02...
> Well, my first recognized cert test today was a bit of a disappointment as
I
> only scored a 770. At least now I know which areas to concentrate on. I
> also got a little balled up early on the first router sim which cost me
some
> time and started to run short toward the end (20 questions remaining and 8
> minutes left...)
>
> There are NO IPX questions at all on this 604-607 exam (I didn't expect
any)
> but... I had at least 3 questions on half/full duplex Ethernet I'm unsure
> of. I know full duplex uses it's own pair to transmit and a separate pair
> to receive thus dramatically reducing collisions but does it eliminate
> collisions altogether? I thought, no.
>
> Secondly, they asked if full duplex incorporated it's own private
collision
> domain. I just had never thought of it termed as a "private" domain.
>
>
| |
| Spam Me No More 2003-10-03, 6:24 pm |
| Hi,
Grey's right. Remember too, full duplex uses the same wires as half. The
difference is the CSMA/CD
protocol is disabled. This means in addition to not expecting
collisions, and having no way to deal with
them, the transmitter does not listen to the receiver at the same time.
This was done in half duplex as
a way to check for collisions...if my own signal is OK then I'm not
being "talked over".
An example of half/full duplex can be seen in a telephone call. Both
sides of the call can send and
receive at the same time. But when they do, both parties notice the
collision (i.e. they talk over each other)
and each waits a random time before beginning again. This is exactly
what CSMA/CD does. Notice also
that one reason cell phones are harder to use than desk phones, is they
lack "sidetone" (the ability to hear
yourself when you talk) the same thing happens in a speaker phone. You
don't know that the other side
is hearing you because you can not hear yourself. This sidetone is
analogous to the receiver of an ethernet
card listening to itself in it's own receiver in half duplex mode.
Using a phone for full duplex can be accomplished if I removed the
sidetone. This is done by analog
modems with some of the tones they produce. Also, for the CCNA exam,
remember that because
the phone network is digital these days (basically ISDN between offices)
the generation of sidetone
and the encoding and decoding of analog speech is done at your local
phone company's switch. (as
well as at the other end of the call) and the voice is digital as it
travels from the origin central office to
the destination central office.
Yes, each port on a switch is its own collision domain. Each VLAN on a
switch is its own broadcast domain.
Good luck on the test!
-Lars
Grey wrote:
>Full duplex has no collisions. It has one transmit pair and one receive
>pair, so the incoming and the outgoing traffic don't collide. For full
>duplex to operate, it has to be connected to a switched port because the
>CSMA/CD mechanism is disabled in full duplex. So, theoretically, there can
>only be two hosts communicating with each other (each using one transmit and
>one receive pair). There's no mechanism for a third host to join the
>conversation because the first two hosts don't expect their transmission to
>be interrupted. I'd say that each host has its own "private" collision
>domain because collisions happen when two hosts start transmitting on the
>same pair at approximately the same time. Since either pair has only one
>transmitting host, there's no other host in the collision domain.
>
>Grey
>
>"Old Garb" <OSS@attnet.net> wrote in message
>news:Srkfb.30155$%h1.18190@sccrnsc02...
>Well, my first recognized cert test today was a bit of a disappointment as I
>only scored a 770. At least now I know which areas to concentrate on. I
>also got a little balled up early on the first router sim which cost me some
>time and started to run short toward the end (20 questions remaining and 8
>minutes left...)
>
>There are NO IPX questions at all on this 604-607 exam (I didn't expect any)
>but... I had at least 3 questions on half/full duplex Ethernet I'm unsure
>of. I know full duplex uses it's own pair to transmit and a separate pair
>to receive thus dramatically reducing collisions but does it eliminate
>collisions altogether? I thought, no.
>
>Secondly, they asked if full duplex incorporated it's own private collision
>domain. I just had never thought of it termed as a "private" domain.
>
>
>
>
| |
| Old Garb 2003-10-03, 6:24 pm |
|
"Scooby" <mmscooby1@removeme.earthlink.net> wrote in message =
news:I9lfb.18489$f11.16963@newsread1.news.atl.earthlink.net...
> Yea, what Grey said. Right on.
>=20
> For more clarification... All hosts use 1 pair for sending and 1 =
pair for
> receiving (not just in full duplex). But, where there are only two =
hosts
> (you to another device), they can both talk and receive at the same =
time.
> But, if there are multiple hosts connected through a hub, then each =
person
> talking will require everyone else to listen. If someone else starts
> talking during that time, then everyone will be required to listen to =
two
> conversations at once and they doesn't work (at least not =
electronically).
> You can talk and listen, you just can't listen to two people at the =
same
> time. That is where collisions come in - they say "back off, someone =
is
> already talking".
>=20
> So, if you are in a shared environment, only one person can talk at a =
time.
> Meaning that if someone is talking to me, I'm not going to start =
talking
> since it will cause collisions. When you know that there are only two =
of
> you, you can go full duplex and you can both talk at the same time.
Well shoot. Now I'm more confused than ever. I've been working with =
hubs and switches for many years without an in-depth knowledge of what =
was going on behind the scenes but I knew switches offered higher =
performance than hubs (in most cases).
Switches are still layer-2 devices (most of them, that is...) and all =
ports (non-VLAN) still share a common broadcast domain, right? Each =
port represents it's own unique collision domain but a broadcast packet =
or series of broadcast packets would still be forwarded to ALL the =
switch ports (except, of course, the one it rode in on). So in a =
switched environment running full duplex there are NEVER any collisions =
due to broadcasts?
=20
> "Grey" <bbb@ccc.com> wrote in message
> news:19-dnRAgI9kqROCiU-KYvQ@comcast.com...
> > Full duplex has no collisions. It has one transmit pair and one =
receive
> > pair, so the incoming and the outgoing traffic don't collide. For =
full
> > duplex to operate, it has to be connected to a switched port because =
the
> > CSMA/CD mechanism is disabled in full duplex. So, theoretically, =
there can
> > only be two hosts communicating with each other (each using one =
transmit
> and
> > one receive pair). There's no mechanism for a third host to join the
> > conversation because the first two hosts don't expect their =
transmission
> to
> > be interrupted. I'd say that each host has its own "private" =
collision
> > domain because collisions happen when two hosts start transmitting =
on the
> > same pair at approximately the same time. Since either pair has only =
one
> > transmitting host, there's no other host in the collision domain.
> >
> > Grey
> >
> > "Old Garb" <OSS@attnet.net> wrote in message
> > news:Srkfb.30155$%h1.18190@sccrnsc02...
> > Well, my first recognized cert test today was a bit of a =
disappointment as
> I
> > only scored a 770. At least now I know which areas to concentrate =
on. I
> > also got a little balled up early on the first router sim which cost =
me
> some
> > time and started to run short toward the end (20 questions remaining =
and 8
> > minutes left...)
> >
> > There are NO IPX questions at all on this 604-607 exam (I didn't =
expect
> any)
> > but... I had at least 3 questions on half/full duplex Ethernet I'm =
unsure
> > of. I know full duplex uses it's own pair to transmit and a =
separate pair
> > to receive thus dramatically reducing collisions but does it =
eliminate
> > collisions altogether? I thought, no.
> >
> > Secondly, they asked if full duplex incorporated it's own private
> collision
> > domain. I just had never thought of it termed as a "private" =
domain.
> >
> >
>=20
>
| |
| Scooby 2003-10-03, 7:24 pm |
|
"Old Garb" <OSS@attnet.net> wrote in message
news:SSlfb.210153$mp.130222@rwcrnsc51.ops.asp.att.net...
"Scooby" <mmscooby1@removeme.earthlink.net> wrote in message
news:I9lfb.18489$f11.16963@newsread1.news.atl.earthlink.net...
>> Yea, what Grey said. Right on.
>>
>> For more clarification... All hosts use 1 pair for sending and 1 pair
for
>> receiving (not just in full duplex). But, where there are only two hosts
>> (you to another device), they can both talk and receive at the same time.
>> But, if there are multiple hosts connected through a hub, then each
person
>> talking will require everyone else to listen. If someone else starts
>> talking during that time, then everyone will be required to listen to two
>> conversations at once and they doesn't work (at least not
electronically).
>> You can talk and listen, you just can't listen to two people at the same
>> time. That is where collisions come in - they say "back off, someone is
>> already talking".
>>
>> So, if you are in a shared environment, only one person can talk at a
time.
>> Meaning that if someone is talking to me, I'm not going to start talking
>> since it will cause collisions. When you know that there are only two of
>> you, you can go full duplex and you can both talk at the same time.
>>
>Well shoot. Now I'm more confused than ever. I've been working with hubs
and switches for many years without an in-depth knowledge of what was going
on behind the scenes
>but I knew switches offered higher performance than hubs (in most cases).
>Switches are still layer-2 devices (most of them, that is...) and all ports
(non-VLAN) still share a common broadcast domain, right? Each port
represents it's own unique collision >domain but a broadcast packet or
series of broadcast packets would still be forwarded to ALL the switch ports
(except, of course, the one it rode in on). So in a switched >environment
running full duplex there are NEVER any collisions due to broadcasts?
Correct, same broadcast domain, different collision domains. Maybe I threw
you off by saying that it can't listen to multiple conversations at the same
time. I mean at the exact same moment. There can be traffic coming in to
you from different locations. Part of the what a switch does is queue these
packets up for you. So, you could get 1 packet from host A, one packet from
host B, one from host C and then do it all over again. A switch port won't
try to forward you three packets at the same time, it will queue them.
It is correct to say that collisions NEVER happen in a full duplex
environment. So, if that is correct, why do they still call them collision
domains in a switched environment? I suppose because a switch can still run
at half duplex and can still suffer collisions with the ports doing so. You
can bet if you look at port statistics and there are collisions, that the
port is (or at least was at some point) running at half duplex.
Switches don't only offer better performance than hubs, but much better
security by better controlling who gets to see what packets.
<snip>
| |
| Old Garb 2003-10-03, 7:24 pm |
|
"Scooby" <mmscooby1@removeme.earthlink.net> wrote in message =
news Vmfb.18623$f11.3746@newsread1.news.atl.earthlink.net...
>=20
> "Old Garb" <OSS@attnet.net> wrote in message
> news:SSlfb.210153$mp.130222@rwcrnsc51.ops.asp.att.net...
>=20
> "Scooby" <mmscooby1@removeme.earthlink.net> wrote in message
> news:I9lfb.18489$f11.16963@newsread1.news.atl.earthlink.net...
> >> Yea, what Grey said. Right on.
> >>
> >> For more clarification... All hosts use 1 pair for sending and 1 =
pair
> for
> >> receiving (not just in full duplex). But, where there are only two =
hosts
> >> (you to another device), they can both talk and receive at the same =
time. [colo
r=darkred]
> >> But, if there are multiple hosts connected through a hub, then each
> person
> >> talking will require everyone else to listen. If someone else =[/color]
starts
> >> talking during that time, then everyone will be required to listen =
to two[c
olor=darkred]
> >> conversations at once and they doesn't work (at least not
> electronically).
> >> You can talk and listen, you just can't listen to two people at the =[/color]
same[
color=darkred]
> >> time. That is where collisions come in - they say "back off, =[/color]
someone is[co
lor=darkred]
> >> already talking".
> >>
> >> So, if you are in a shared environment, only one person can talk at =[/color]
a
> time.
> >> Meaning that if someone is talking to me, I'm not going to start =
talking
> >> since it will cause collisions. When you know that there are only =
two of[co
lor=darkred]
> >> you, you can go full duplex and you can both talk at the same time.
> >>
>=20
>=20
> >Well shoot. Now I'm more confused than ever. I've been working with =[/color]
hubs
> and switches for many years without an in-depth knowledge of what was =
going
> on behind the scenes
> >but I knew switches offered higher performance than hubs (in most =
cases).
>=20
> >Switches are still layer-2 devices (most of them, that is...) and all =
ports
> (non-VLAN) still share a common broadcast domain, right? Each port
> represents it's own unique collision >domain but a broadcast packet or
> series of broadcast packets would still be forwarded to ALL the switch =
ports
> (except, of course, the one it rode in on). So in a switched =
>environment
> running full duplex there are NEVER any collisions due to broadcasts?
>=20
>=20
> Correct, same broadcast domain, different collision domains. Maybe I =
threw
> you off by saying that it can't listen to multiple conversations at =
the same
> time. I mean at the exact same moment. There can be traffic coming =
in to
> you from different locations. Part of the what a switch does is queue =
these
> packets up for you. So, you could get 1 packet from host A, one =
packet from
> host B, one from host C and then do it all over again. A switch port =
won't
> try to forward you three packets at the same time, it will queue them.
>=20
> It is correct to say that collisions NEVER happen in a full duplex
> environment. So, if that is correct, why do they still call them =
collision
> domains in a switched environment? I suppose because a switch can =
still run
> at half duplex and can still suffer collisions with the ports doing =
so. You
> can bet if you look at port statistics and there are collisions, that =
the
> port is (or at least was at some point) running at half duplex.
>=20
> Switches don't only offer better performance than hubs, but much =
better
> security by better controlling who gets to see what packets.
>=20
> <snip>
Got it. That queing (buffering) thing. Mucho grass y'all. Hope to =
retest in two weeks.
| |
| Alan Spicer 2003-10-07, 1:26 am |
|
>Well shoot. Now I'm more confused than ever. I've been working with hubs
and switches for many years without an in->depth knowledge of what was going
on behind the scenes but I knew switches offered higher performance than
hubs (in most >cases).
>Switches are still layer-2 devices (most of them, that is...) and all ports
(non-VLAN) still share a common broadcast domain, >right? Each port
represents it's own unique collision domain but a broadcast packet or series
of broadcast packets would >still be forwarded to ALL the switch ports
(except, of course, the one it rode in on). So in a switched environment
running >full duplex there are NEVER any collisions due to broadcasts?
* Switch is a multi-port transparent bridge. Yes it is still layer-2 device
and yes all of these ports still share the same broadcast domain. But not
the same collision domain. The broadcasts don't have anything to collide
with ... the broadcasting is all done within the switch fabric or core of
the switch. The layer two hosts on the switch ports are still on dedicated
segments so collisions just do not occur. Every device on a switch port will
receive any broadcasts ... but only on the receive side. There is nothing
else there talking to a device except for the switch port. The host talks on
the transmit pair. Nothing could collision.
Now if you've got a big broadcast storm problem ... that needs to be sorted
out by troubleshooting the cause. Possibly VLAN's would need to be
implemented to seperated broadcast domains. Or use a layer-3 device.
If an ethernet broadcast or multicast frame comes into a switch, the switch
will forward the frame to all switch ports until a MAC host acknowledges it
.... and then switch will have that MAC address in the forwarding table and
what port it is on.
If someone is playing multicast game or application, even only 2 hosts, and
every port will be bothered by these multicast packets.
---
Alan Spicer (a_spicerNOSPAM@bellsouth.net)
http://aspicer.homelinux.net/
Systems and Network Administration,
and Telecommunications
| |
| Alan Spicer 2003-10-07, 1:26 am |
| "Scooby" <mmscooby1@removeme.earthlink.net> wrote in message
news:I9lfb.18489$f11.16963@newsread1.news.atl.earthlink.net...
> Switches don't only offer better performance than hubs, but much better
> security by better controlling who gets to see what packets.
Some switches can be tricked into mirroring their ports ... I think in some
cases this is a hack. Cisco switches can be configured by the administrator
to do Port Spanning, typically for Intrusion Detection monitoring. But if a
bad person could cause this kind of port mirroring (and I remember reading
before that it has been done) then there goes the security aspect.
---
Alan Spicer (a_spicerNOSPAM@bellsouth.net)
http://aspicer.homelinux.net/
Systems and Network Administration,
and Telecommunications
| |
| Alan Spicer 2003-10-07, 1:26 am |
| * In the following post please ignore the part "until a MAC host
acknowledges it
.... and then switch will have that MAC address in the forwarding table and
what port it is on."
I had meant to edit it out as I was thinking of two things at the same time
and I forgot to finish the editing of those lines. I was thinking of both
cases of:
1.) An unknown host or MAC id on an unknown switch port, in which case the
switch floods the frame until the host answers telling what port it is on.
2.) Multicast case and all broadcasts which go to all ports always. If the
broadcast was looking for a particular host the broadcast would stop once it
was found. (does that sound right?)
>
> If an ethernet broadcast or multicast frame comes into a switch, the
switch
> will forward the frame to all switch ports until a MAC host acknowledges
it
> ... and then switch will have that MAC address in the forwarding table and
> what port it is on.
>
> If someone is playing multicast game or application, even only 2 hosts,
and
> every port will be bothered by these multicast packets.
>
>
>
> ---
> Alan Spicer (a_spicerNOSPAM@bellsouth.net)
> http://aspicer.homelinux.net/
> Systems and Network Administration,
> and Telecommunications
>
>
| |
| Scooby 2003-10-07, 9:24 am |
| No, switches can't be "tricked" into mirroring ports. They can be
configured to do so, but you need to have the correct access to the switch.
Your network admin is responsible for enforcing security, not breaking it.
The port mirroring is a way for the net admin to monitor traffic without
having to change the switch out for a hub. Actually, it is easier for a
hacker to just put a hub between the target and the switch and they can
monitor all they want.
Anyway, I didn't say that it was 100% secure, nothing is. Everything has a
weakness. But, this is much more secure than a hub.
"Alan Spicer" <a_spicer@bellsouth.net> wrote in message
news:nErgb.24465$wC1.5122@bignews3.bellsouth.net...
> "Scooby" <mmscooby1@removeme.earthlink.net> wrote in message
> news:I9lfb.18489$f11.16963@newsread1.news.atl.earthlink.net...
>
> > Switches don't only offer better performance than hubs, but much better
> > security by better controlling who gets to see what packets.
>
> Some switches can be tricked into mirroring their ports ... I think in
some
> cases this is a hack. Cisco switches can be configured by the
administrator
> to do Port Spanning, typically for Intrusion Detection monitoring. But if
a
> bad person could cause this kind of port mirroring (and I remember reading
> before that it has been done) then there goes the security aspect.
>
> ---
> Alan Spicer (a_spicerNOSPAM@bellsouth.net)
> http://aspicer.homelinux.net/
> Systems and Network Administration,
> and Telecommunications
>
>
| |
| Scooby 2003-10-07, 10:24 am |
| "Alan Spicer" <a_spicer@bellsouth.net> wrote in message
news:OJrgb.24470$wC1.9870@bignews3.bellsouth.net...
> * In the following post please ignore the part "until a MAC host
> acknowledges it
> ... and then switch will have that MAC address in the forwarding table and
> what port it is on."
>
> I had meant to edit it out as I was thinking of two things at the same
time
> and I forgot to finish the editing of those lines. I was thinking of both
> cases of:
>
> 1.) An unknown host or MAC id on an unknown switch port, in which case the
> switch floods the frame until the host answers telling what port it is on.
Nope, it sends it out once to all ports (except the incoming port). If
nobody answers, the switch does not care.
>
> 2.) Multicast case and all broadcasts which go to all ports always. If the
> broadcast was looking for a particular host the broadcast would stop once
it
> was found. (does that sound right?)
>
Broadcasts go out all ports (except the incoming port). They are not for
one particular host, that is why they are broadcasts. They are for anyone
who cares. And, the switch does not keep sending the same packet - it only
sends it once.
> >
> > If an ethernet broadcast or multicast frame comes into a switch, the
> switch
> > will forward the frame to all switch ports until a MAC host acknowledges
> it
> > ... and then switch will have that MAC address in the forwarding table
and
> > what port it is on.
> >
> > If someone is playing multicast game or application, even only 2 hosts,
> and
> > every port will be bothered by these multicast packets.
> >
> >
> >
> > ---
> > Alan Spicer (a_spicerNOSPAM@bellsouth.net)
> > http://aspicer.homelinux.net/
> > Systems and Network Administration,
> > and Telecommunications
> >
> >
>
>
| |
| Lisa Duchovny 2003-10-08, 4:25 am |
| Hi,
Do not get disappointed. There is always a next time! Believe in
yourself and the saying, it always gets better!
"Alan Spicer" <a_spicer@bellsouth.net> wrote in message news:<OJrgb.24470$wC1.9870@bignews3.bellsouth.net>...
> * In the following post please ignore the part "until a MAC host
> acknowledges it
> ... and then switch will have that MAC address in the forwarding table and
> what port it is on."
>
> I had meant to edit it out as I was thinking of two things at the same time
> and I forgot to finish the editing of those lines. I was thinking of both
> cases of:
>
> 1.) An unknown host or MAC id on an unknown switch port, in which case the
> switch floods the frame until the host answers telling what port it is on.
>
> 2.) Multicast case and all broadcasts which go to all ports always. If the
> broadcast was looking for a particular host the broadcast would stop once it
> was found. (does that sound right?)
>
> >
> > If an ethernet broadcast or multicast frame comes into a switch, the
> switch
> > will forward the frame to all switch ports until a MAC host acknowledges
> it
> > ... and then switch will have that MAC address in the forwarding table and
> > what port it is on.
> >
> > If someone is playing multicast game or application, even only 2 hosts,
> and
> > every port will be bothered by these multicast packets.
> >
> >
> >
> > ---
> > Alan Spicer (a_spicerNOSPAM@bellsouth.net)
> > http://aspicer.homelinux.net/
> > Systems and Network Administration,
> > and Telecommunications
> >
> >
| |
| Al Dykes 2003-10-25, 1:25 pm |
| In article <19-dnRAgI9kqROCiU-KYvQ@comcast.com>, Grey <bbb@ccc.com> wrote:
>Full duplex has no collisions. It has one transmit pair and one receive
>pair, so the incoming and the outgoing traffic don't collide. For full
>duplex to operate, it has to be connected to a switched port because the
>CSMA/CD mechanism is disabled in full duplex. So, theoretically, there can
>only be two hosts communicating with each other (each using one transmit and
>one receive pair). There's no mechanism for a third host to join the
>conversation because the first two hosts don't expect their transmission to
>be interrupted. I'd say that each host has its own "private" collision
>domain because collisions happen when two hosts start transmitting on the
>same pair at approximately the same time. Since either pair has only one
>transmitting host, there's no other host in the collision domain.
>
>Grey
>
What happens if the sender is pushing (or trying to push) packets
faster than the receiving machine (or the switch) can handle ? I always
figured that a colision was simulated to make the sender back off..
>"Old Garb" <OSS@attnet.net> wrote in message
>news:Srkfb.30155$%h1.18190@sccrnsc02...
>Well, my first recognized cert test today was a bit of a disappointment as I
>only scored a 770. At least now I know which areas to concentrate on. I
>also got a little balled up early on the first router sim which cost me some
>time and started to run short toward the end (20 questions remaining and 8
>minutes left...)
>
>There are NO IPX questions at all on this 604-607 exam (I didn't expect any)
>but... I had at least 3 questions on half/full duplex Ethernet I'm unsure
>of. I know full duplex uses it's own pair to transmit and a separate pair
>to receive thus dramatically reducing collisions but does it eliminate
>collisions altogether? I thought, no.
>
>Secondly, they asked if full duplex incorporated it's own private collision
>domain. I just had never thought of it termed as a "private" domain.
>
>
--
Al Dykes
-----------
adykes@panix.com
| |
|
| > What happens if the sender is pushing (or trying to push) packets
> faster than the receiving machine (or the switch) can handle ? I always
> figured that a colision was simulated to make the sender back off..
>
If the hardware buffers on the receiver fills faster than can be emptied,
then frames will be dropped. Then it is up to the upper layers to recover.
Hope this helps,
dg
|
|
|
|
|