At 02:28 PM 1/24/01, David L. Blair wrote:
>I know the reason for two of three.
>
>1) Why is the Ethernet minimum frame size of 64 bytes?
>
>Ethernet timing is based upon "bit time" the time a bit takes to travel the
>distance of an maximum Ethernet segment length 100m or 328 ft.
Don't get me started again! &;-) The maximum Ethernet topology is usually
much bigger than 100 meters on a shared Ethernet network with hubs, shared
coax cable, etc. See my chart here:
http://www.priscilla.com/enetscales.htm. Also, keep in mind that what
matters is the round-trip time.
The physical size of the network topology is limited so that a station is
still sending the minimum size frame (64 bytes) if a collision occurs and
reflects back to the sender. If the sender were not still sending, the
frame would have to be retransmitted by a higher layer. For a higher layer
to notice that the frame needs retransmitting takes much more time.
Ethernet retransmissions occur within nanoseconds usually. That's the key
to understanding Ethernet parameters. You have to think about WHY, not just
memorize rules and jargon about "bit times."
> The minimum
>frame size 64 bytes equals 512 bits or 512 bit times to travel the wire.
Make that "travel the network segment and back," or "travel the collision
domain and back" or something other than "travel the wire" since wire could
just mean the cable from the station to the hub.
>This has to do with a host on one end of the wire listening to use the wire
>and another host on the opposite end transmitting. Note: This does not
>include the Preamble.
>2) Why is the Ethernet maximum frame size of 1518 bytes(not 1526)?
>
>I guess for a similar reason.
It's not for the same reason. The minimum frame is used to ensure correct
collision detection and retransmission. The maximum frame size is used to
make sure there aren't any bandwidth hogs, as you describe below.
Where did 1526 come from? For some reason, nobody ever counts the preamble,
if that's what the thinking was.
>Remember most Ethernet implementations are
>share media. So everyone host is contending for access to the wire. That
>is why Ethernet is contention based and Token Ring & FDDI are deterministic.
>A packet to large would actually allow less hosts per segment to transmit
>their data in a timely fashion.
>
>3) Why is the ATM frame size 53 bytes?
>
>I do know exactly why 53 bytes was picked. I do know why a small frame size
>was picked. A ATM packet has two parts: Address and Payload. The Address
>is 5 bytes long and the Payload is 48 bytes longs. ATM was designed as a
>Multi-service access medium, i.e. to handle Data, Voice, and Multimedia
>content. The "Holy Grail" of voice transmission is delay and jitter. When
>the Delay in voice transmission is more than 250ms, humans start talking
>before the other person has finished. Resulting in a garbled conversion.
>It takes less time to fill small frames with data. All the frames are the
>same size so the transmission time, for a given network, is constant. QoS,
>Quality of Service, features are implemented in ATM to guarantee delivery of
>time sensitive frames like voice.
Sounds good to me. Apply that level of thinking to your Ethernet
understanding too and you'll be in good shape. &:-)
Priscilla
>I hoe this helps. A good Ethernet book with some ATM and FDDI information
>is Switched, Fast, and Gigabit Ethernet ISBN: 1-57870-073-6
>
>-dlb
>
>
>"azhar mumtaz" <[EMAIL PROTECTED]> wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > Hello Guys:
> > What i want to know is why there is a limit that ethernet frame size
>cannot be
> > less than 64 bytes and more than 1526 bytes. I know that this is how
>ethernet
> > should be understand but whats the logic behind it. Similarly why we are
> > limiting cell size of ATM to 53 bytes.
> > Regards
> > Azhar Soomro
> >
> > ____________________________________________________________________
> > Get free email and a permanent address at http://www.netaddress.com/?N=1
> >
> > _________________________________
> > 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]