Date: Wed, 7 Apr 1999 17:16:25 +0200 From: Ladavac Marino <mladavac@metropolitan.at> To: 'Walter Hafner' <hafner@informatik.tu-muenchen.de>, Joe Abley <jabley@clear.co.nz> Cc: freebsd-hackers@freebsd.org Subject: RE: IP Type of service (FTP proxy in German c`t) Message-ID: <97A8CA5BF490D211A94F0000F6C2E55D09758E@s-lmh-wi-900.corpnet.at>
next in thread | raw e-mail | index | archive | help
> -----Original Message----- > From: Walter Hafner [SMTP:hafner@informatik.tu-muenchen.de] > Sent: Wednesday, April 07, 1999 4:57 PM > To: Joe Abley > Cc: freebsd-hackers@freebsd.org; hafner@informatik.tu-muenchen.de > Subject: Re: IP Type of service (FTP proxy in German c`t) >=20 > First of all: >=20 > A couple of people mailed me directly and told me, that I'm probably > the=20 > victim of an April fools joke of the c't (The article was in the > April's=20 > issue). I had the suspection myself, therefore I tested the program > before I posted. >=20 > As you can see below, I really got speed improvements. If this is = just > coincidence, well, go on, laugh at me. :-) >=20 > (Ok. You can stop now.) >=20 > If not, then I see a problem. >=20 >=20 >=20 > Joe Abley <jabley@clear.co.nz> writes: >=20 > > If the intention is to accelerate downloads, then I don't think = this > > proxy is going to do much. It might promote the treatment of = packets > outbound > > from the client if the rfc1349 precedence is set to 7, but it can > have > > no control over the return packets -- and that's where all the data > is > > on a download (which is what most people do). [ML] Unless the proxy always tries to use passive mode thus opening BOTH connections if the server supports it. The observed results would seem to agree. > I had a look on 1349, as soon as I read the article. The article > states, that "normally" all TOS and precedence settings from incoming > packets are simply copied to the outgoing packets (which makes > sense). That's the reason I checked the RFC. Unfortunately I didn't > find > it mentioned. RCF1195 doesn't say much about it either. >=20 > I can't verify the assumption in detail (I=B4m no expert), but in a = few > quick tests I watched this or a similar behaviour: >=20 [ML] One should really try to tcpdump the connections. [measurements deleted] Sounds like trouble /Marino To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?97A8CA5BF490D211A94F0000F6C2E55D09758E>