It should say 1432 bytes left over for payload once you subtract the
following bytes from the maximum FDDI frame size of 4500 bytes:
20 bytes for FDDI
8 bytes for LLC/SNAP
20 bytes for IP
20 bytes for TCP
This would be counting the preamble, starting delimiter, frame control and
FCS for the FDDI frame, but not counting the ending delimiter or frame
status symbols. Also, FDDI doesn't really have a maximum frame size. It has
a token holding timer which results in approximately 4500 bytes being the
maximum a single station can send.
And the following words from the white paper should be tweaked a bit:
"Since each frame would add a packet, you would add the framing, LLC/SNAP,
IP, and TCP overhead on each packet. Adding two additional frames adds 144
bytes of overhead at all affected layers, a small amount of overhead when
compared with the overhead you would incur in fragmenting on WANs."
The first frame would have 68 bytes of overhead: FDDI, LLC, SNAP, IP and TCP.
The two fragments would have only 48 bytes of overhead. The TCP header is
not replicated in an IP fragment. IP fragments just have an IP header and
data. Still a small amount of overhead, as the paper says.
Let me know if I've said anything wrong. Thanks.
Priscilla
At 10:29 AM 7/13/00, Phil Barker wrote:
>Hi group,
> Just about finished this excellent paper but have
>come across the following issue.
>
>under the heading of "IP Fragmentation"
>
> >>> SNIP
>On a FDDI medium with a 4500 byte maximum frame size,
>you have 4436 bytes available for payload in each
>frame. This payload includes 8 bytes of LLC/SNAP, 10
>bytes of IP, and 20 bytes of TCP. There are also 20
>bytes of FDDI frame overhead.
>
>In the ideal case, there is a single transmitter
>sending sequential frames. The theoretical efficiency
>is 98.5%. With a 1500 byte frame payload limit, the
>transmitter would need to create three frames to carry
>the same information. Since each frame would add a
>packet, you would add the framing, LLC/SNAP, IP, and
>TCP overhead on each packet. Adding two additional
>frames adds 144 bytes of overhead at all affected
>layers, a small amount of overhead when compared with
>the overhead you would incur in fragmenting on WANs.
>
> >>> END SNIP
>
> From the first paragraph.
>I agree with 8 bytes of LLC SNAP.
>and I agree with 20 bytes of TCP header.
>But I think 10 bytes of IP should also be 20 ?
>And the overall sentence should read that the
>remaining header i.e 4500-4436 = 64 consists of
>(8 + 8 + 8 + 20 + 20) = 64
>(Dest mac + Source mac + LLC/SNAP + IP + TCP)
>
>Can anyone confirm ?
>
>Phil.
>
>
>____________________________________________________________
>Do You Yahoo!?
>Get your free @yahoo.co.uk address at http://mail.yahoo.co.uk
>or your free @yahoo.ie address at http://mail.yahoo.ie
>
>___________________________________
>UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
>FAQ, list archives, and subscription info: http://www.groupstudy.com
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
___________________________________
UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
FAQ, list archives, and subscription info: http://www.groupstudy.com
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
