Very well said Priscilla, more or less what I was trying to say from memory.
Also included in this is that when a collision is detected a jam is sent on
the wire and the back off mechanism comes into play, should another
collision occur, then another back off of a longer period takes place, and
so on. After 4 (If I remember correctly) back offs the packet is dropped
and left to a higher layer protocol to retransmit. You are correct that a
late collision indicates a collision that is past the preamble and should
never happen in a properly designed and specified Ethernet segment, however
when they do occur it is most likely a cable that is beyond length in spec.
While we are on the topic I am often asked what a runt is, simply
put it is the fragments that result in collisions on an Ethernet segment, a
somewhat normal condition. With Store and Forward switching runts will be
eliminated from the wire while with cut-through switching they can be
propagated. This being said, even though Cut-Through can be faster on a
lightly loaded network store and forward can provide for higher throughput
on a more saturated network due to this fact.
Also I am asked what giants are. They are the result of an Ethernet
frame being larger than the IEEE limit of 1524 (Or 1518 depending on who
your talking to) Bytes. This can be from a few things, the most important
being VLAN tagging. Other sources are mis-configured stations on the wire
or NIC's that are spewing garbage on the wire.
For anyone really interested in a GREAT sight covering these items
the original writer of the CNX certification has an excellent web site with
all these goodies on it. It is www.optimized.com go there and check out the
Technical Compendium link.
Merry Christmas to all,
Shawn
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
Priscilla Oppenheimer
Sent: Monday, December 25, 2000 7:47 PM
To: Bowen, Shawn; Li Song; [EMAIL PROTECTED]
Subject: RE: is this statement true ??
At 01:07 PM 12/25/00, Bowen, Shawn wrote:
>According to IEEE NO, 100 Meters is the max cable distance for Half or Full
>100MB Ethernet over TP. In reality, Yes it will extend the range, The
>reason why is that at full duplex you can not have collisions, and
>collisions are the main reason for the distance limitation (Cross Talk
comes
>into play as well). The reason behind this is that in the original IEEE
>spec the distance limitation was set so that a single 64Byte packet (the
>smallest) could be transmitted down the line and would collide with another
>packet before the 64Byte packet header had been completely transmitted,
when
Minor correction: The distance limitation is defined so that if a station
is transmitting a minimum-size frame (64 bytes) and a collision occurs at
the other end of the network, the collision will reflect back to the sender
while the sender is still sending. If this didn't happen, the sender would
have stopped monitoring for a collision with its transmission, and would
not automatically retry. An upper layer would have to retransmit, which
takes a lot longer.
>this does not happen properly you see late collisions, these indicate a
>collision past the preamble header of the packet and indicate a cable
length
>that is to long.
A late collision is one that happens past the preamble and past the first
64 bytes of the frame. A collision within the first 64 bytes is legal,
normal, and not late.
Priscilla
>If you need to even go close to the 100 Meter mark you
>should consider 100BaseFX or similar.
>
>Shawn
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Li
>Song
>Sent: Monday, December 25, 2000 4:33 AM
>To: [EMAIL PROTECTED]
>Subject: is this statement true ??
>
>"full-duplex can be used over longer distance than
>half-duplex" ??
>what 's your opinion ??
>
>
>_________________________________
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
>_________________________________
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
_________________________________
FAQ, list archives, and subscription info:
http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
