unidirectional file transfer over IP
Innocenti Maresin
There are many protocols for file transfer over IP (FTP, HTTP, NFS, SMB...)
but these protocols generate a lot of reverse traffic,
FTP and HTTP because of use of TCP (which sends many acknowledgements),
others because of its "query-answer" semantic.
I need a protocol for file transfer which transmits packets mostly unidirectional,
from the file source to the consumer, using really few queries and acknowledgements.
Perhaps, TFTP is such protocol (it uses UDP),
but I never used TFTP so I don't know how to transfer a big files with it.
The problem is that my outcoming traffic is absolutely free, but incoming isn't.
Thanks in advance.
--
qq~~~~\
/ /\ \
\ /_/ /
\____/
Barry Margolin
Innocenti Maresin <av...@comtv.ru> wrote:
>There are many protocols for file transfer over IP (FTP, HTTP, NFS, SMB...)
>but these protocols generate a lot of reverse traffic,
>FTP and HTTP because of use of TCP (which sends many acknowledgements),
FTP should normally send an ACK for every 2 data segments. If the MSS is
around 1500, that's 40 bytes of return traffic for every 3000 bytes of
data. Is that really so bad?
>others because of its "query-answer" semantic.
>I need a protocol for file transfer which transmits packets mostly
>unidirectional,
>from the file source to the consumer, using really few queries and
>acknowledgements.
>Perhaps, TFTP is such protocol (it uses UDP),
No, it sends an acknowledgement packet for every data packet (which is
normally 512 bytes), so it generates more return traffic than FTP typically
does.
>but I never used TFTP so I don't know how to transfer a big files with it.
You transfer big files the same way that you transfer small ones.
--
Barry Margolin, barry.m...@level3.com
Genuity Managed Services, a Level(3) Company, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.
Glen Herrmannsfeldt
"Barry Margolin" <barry.m...@level3.com> wrote in message
news:2fPta.3$_74...@paloalto-snr1.gtei.net...
> In article <3EB76513...@comtv.ru>,
> Innocenti Maresin <av...@comtv.ru> wrote:
> >There are many protocols for file transfer over IP (FTP, HTTP, NFS,
SMB...)
> >but these protocols generate a lot of reverse traffic,
> >FTP and HTTP because of use of TCP (which sends many acknowledgements),
(snip)> >Perhaps, TFTP is such protocol (it uses UDP),
> No, it sends an acknowledgement packet for every data packet (which is
> normally 512 bytes), so it generates more return traffic than FTP
typically
> does.
>
> >but I never used TFTP so I don't know how to transfer a big files with
it.
>
> You transfer big files the same way that you transfer small ones.
The only one I ever knew to use TFTP was booting diskless machines. It is
simpler, so easier to fit into boot roms. What they did was use TFTP to
load the loading program, which then used a different protocol to load the
OS. Sun used NFS for the second step, traditionally also UDP, but much
faster than TFTP.
I believe TFTP acks every packet, and doesn't send one until the previous
ack arrives. Sun kept a running display on the screen, so you could see
how fast it wasn't.
-- glen
Innocenti Maresin
What OS and FTP-client I shoud use to minimize return traffic?
Some people says that return traffic is up to 10% of data.
It the "TCP window size" critical only in the client box, or also in the server?
What other settings are relevant?
Barry Margolin wrote in news:2fPta.3$_74....@paloalto-snr1.gtei.net
> FTP should normally send an ACK for every 2 data segments. If the MSS is
> around 1500, that's 40 bytes of return traffic for every 3000 bytes of
> data. Is that really so bad?
_This_ isn't so bad. But there is a risk of use of FTP-clients or TCP stack software
which will flood the server with ACKs much more than 1 packet to 3KB of data.
So opening a guest access for all Internet is very dangerous.
Barry Margolin
Innocenti Maresin <av...@comtv.ru> wrote:
>OK, thanks, I'll try to ask in another way.
>What OS and FTP-client I shoud use to minimize return traffic?
>Some people says that return traffic is up to 10% of data.
>It the "TCP window size" critical only in the client box, or also in the server?
>What other settings are relevant?
Window size has little to do with it. Segments are normally sized based on
the MTU, which is usually 1000-1500 bytes. Acknowledgements are normally
sent for every 2 segments.
>Barry Margolin wrote in news:2fPta.3$_74....@paloalto-snr1.gtei.net
>
>> FTP should normally send an ACK for every 2 data segments. If the MSS is
>> around 1500, that's 40 bytes of return traffic for every 3000 bytes of
>> data. Is that really so bad?
>
>_This_ isn't so bad. But there is a risk of use of FTP-clients or TCP
>stack software
>which will flood the server with ACKs much more than 1 packet to 3KB of data.
>So opening a guest access for all Internet is very dangerous.
At the worse they'll ack every segment instead of every other segment, so
it will be 2% return traffic instead of 1%.
Kevin D. Quitt
I do have sample code.
--
#include <standard.disclaimer>
_
Kevin D Quitt USA 91387-4454 96.37% of all statistics are made up
Per the FCA, this address may not be added to any commercial mail list