Text 872, 62 rader
Skriven 2007-05-04 07:43:02 av Maurice Kinal (1:261/38.9)
    Kommentar till text 869 av andrew clarke (3:633/267.0)
Ärende: binkp vs ftp
====================
Hey andrew!

May 04 16:59 07, andrew clarke wrote to Maurice Kinal:

 ac> They aren't supposed to.  But FTP isn't designed to transfer FidoNet 
 ac> mail.

Really?  Could have fooled me.  It's been working great for exactly that for
many, may years now.  Much better than anything else, nevermind Fido mailers.

 ac> In particular, there's no reliable mechanism for FTP servers 
 ac> to truncate or delete mail bundles after they've been sent to another 
 ac> node.

That comes as a total surprise to me.  How is it that it can toss this message
that I am replying to as weil as soon to toss this reply?  Not to mention all
the previous echomail (netmail too) over these many, many years.  I have yet to
see anything better.

 ac> Unlike a BinkP client, an FTP client has no way to acknowledge 
 ac> that it received a file.

Really?  Could have fooled me.  Jeez.  You have me stumped here why it seems to
be working.

 ac> BinkP is also considerably easier to configure firewall rules for, 
 ac> because it just uses a single TCP port for 
 ac> communication, unlike FTP.

Sounds good.  Myself I'll stick to the configuration here until I see better.

 MK>> Binkd is about 1/3 as efficient as ftpd at pure file transfers not 
 MK>> counting additional processing it requires.

 ac> I don't know where you're pulling those figures from.

It comes from observation of transfer speeds, cpu usage, etc.  I compare and
contrast.  1/3 is being generous on my part as I didn't factor in the total
processing time (tossing etc) and only the actual transfer of the file(s) from
connection to end of transfer.

 ac> There is not 
 ac> much overhead in the BinkP protocol.

Slow in comparison.

 ac> Unless you're referring to a 
 ac> particular version of BinkD, which is slow for you, for some odd 
 ac> reason.

I have kept up.  Also used a +700M file over a lan to compare and did it many
times to be sure.  1/3 efficiency is a good guesstimate.  Set up some
parameters and I will be happy to share the results with you.  At the moment I
am using vsftpd-2.0.5 and will gladly download and build a binkd to the same
machine specs to ensure fairness.  Also I can assure that the network will
dedicate 100% of it's attention to this experiment.

Life is good,
Maurice

--- Msged/LNX 6.2.0
 * Origin: The Pointy Stick Society XXIX - Only one I in teamwork (1:261/38.9)