| |
PA2AGA > HDDIG 26.11.99 06:34l 229 Lines 7733 Bytes #-9671 (0) @ EU
BID : HD_99_302C
Read: DL6KCF GUEST
Subj: HamDigitalDigest 99/302C
Path: DB0AAB<DB0ZKA<DB0LX<DB0RBS<DB0SEL<DB0ZDF<DB0AIS<DB0ME<DB0QS<DB0ACC<
DB0SM<PI8DAZ<PI8GCB<PI8HGL<PE1NMB<EA7URC<PE0MAR<PI8VNW
Sent: 991126/0130Z @:PI8VNW.#ZH2.NLD.EU #:24878 [HvHolland] FBB7.00g $:HD_99_30
From: PA2AGA@PI8VNW.#ZH2.NLD.EU
To : HDDIG@EU
Received: from pa2aga by pi1hvh with SMTP
id AA24377 ; Fri, 26 Nov 99 00:50:34 UTC
Received: from pa2aga by pa2aga (NET/Mac 2.3.67/7.5.3) with SMTP
id AA00016995 ; Fri, 26 Nov 99 00:52:08 MET
Date: Fri, 26 Nov 99 00:51:36 MET
Message-Id: <hd_99_302C>
From: pa2aga
To: hd_broadcast@pa2aga
Subject: HamDigitalDigest 99/302C
X-BBS-Msg-Type: B
drops packets
> >do to random errors, and not due to congestion.
>
> Right. So just fix that.
>
> Rob
> --
>
+----------------------------------+--------------------------------------+
> | Rob Janssen pe1chl@amsat.org | WWWhome: http://www.pe1chl.demon.nl/
|
> | AMPRnet: rob@pe1chl.ampr.org | AX.25 BBS: PE1CHL@PI8WNO.#UTR.NLD.EU
|
>
+----------------------------------+--------------------------------------+
>.
------------------------------
Date: Sun, 21 Nov 1999 11:36:37 -0800
From: "Cathryn Mataga" <cathryn@junglevision.com>
Subject: JNOS vs TNOS??
Oh, I don't know. I'd say around here, that there are as many people on ip,
as
there are on netrom type stuff. Around here, net connections are cheap, so
everybody is one hop away from a gateway, basically, and on one hop
not many of these issues come up -- really. There's not much of a 'network'
for ip stuff. And the netrom network is practically, too slow for ip stuff.
Really
the hard answer to all of this is to never drop packets -- ever, and then
things
run pretty smoothly.
The thing that keeps ip going, is that converse, seems to finally have the
last bugs
killed in it, and there are actually quite a few people on converse pretty
much 24 hours a day -- from what I've seen. Or at least if you want to chat
ham radio stuff on keyboards, it's better than IRC or any of the internet chat
rooms.
(I'm kind of a jammer magnet, so I tend to stay off of repeaters.)
"Rob" <NoEmail@NoWay.com> wrote in message news:g6U_3.193228$5r2.446122@tor-
nn1.netcom.ca...
> It sounds like IP/TCP is dying and most hams are turning back to NETROM (or
> KNET etc)
>
> Rob
>
> "Rob Janssen" <nomail@pe1chl.demon.nl> wrote in message
> news:slrn83nj06.ue2.nomail@linux.pe1chl.ampr.org...
> > Cathryn Mataga <cathryn@junglevision.com> wrote:
> > >I believe there's an advantage to running NOS over internet based IP
> stacks
> > >like with Linux and WIndows, that you can specify that the retry timers
> backoff
> > >linearly, rather than exponentially.
> >
> > That is actually a disadvantage. It causes congestion collapse, and
> > makes outside observers judge TCP/IP as a bad protocol that jams
> > everything else.
> >
> > >Apparently, this is considered extremely
> > >evil from the internet perspective, but it's a fast kluge to get ip to
> work
> > >a little better when the connection is dropping packets due to noisy
> links--
> > >rather than congestion. (There's a linux Kernel hack, I thought someone
> posted
> > >to do this, though I haven't seen anything like this for Windows.)
> >
> > Actually, when your links drop packets you should not attempt to
> retransmit
> > them from the endpoints, but you should adjust your links (individually)
> > to use a virtual circuit rather than a datagram technique to send the
> > packets over that hop.
> >
> > Most NET and NOS versions support this.
> > My version of NET even allows you to setup automatic selection of VC or
> > Datagram mode based on the independently measured packet loss rate on the
> > connection, separately for each direction.
> > This works very well, without requiring inefficient retransmissions from
> > the endpoints that yield TCP/IP a bad reputation.
> >
> > >It's actually, just a hard situation all around, the more I've looked
> into this, since
> > >my understanding is that TCP/IP doesn't really work that efficiently
> either if the
> > >underlying protocol provides reliable connections, or if the channel
> drops packets
> > >do to random errors, and not due to congestion.
> >
> > Right. So just fix that.
> >
> > Rob
> > --
> >
> +----------------------------------+--------------------------------------+
> > | Rob Janssen pe1chl@amsat.org | WWWhome: http://www.pe1chl.demon.nl/
> |
> > | AMPRnet: rob@pe1chl.ampr.org | AX.25 BBS: PE1CHL@PI8WNO.#UTR.NLD.EU
> |
> >
> +----------------------------------+--------------------------------------+
>
>
>.
------------------------------
Date: Wed, 24 Nov 1999 15:11:47 -0500
From: "Rob" <NoEmail@NoWay.com>
Subject: JNOS vs TNOS??
Well I am just learning NOS -- Read NOSIntro and understodd 99% of it.
I will play with JNOS -- Sounds like it is still supported. And I have an
old 386 collecting dust that would be perfect for JNOS.
When you say CONVERSE I assume it is different that the CONVERSE mode in
AX.25.
Rob
"Cathryn Mataga" <cathryn@junglevision.com> wrote in message
news:Z9X_3.1372$LK3.54900@nuq-read.news.verio.net...
> Oh, I don't know. I'd say around here, that there are as many people on
ip, as
> there are on netrom type stuff. Around here, net connections are cheap,
so
> everybody is one hop away from a gateway, basically, and on one hop
> not many of these issues come up -- really. There's not much of a
'network'
> for ip stuff. And the netrom network is practically, too slow for ip
stuff. Really
> the hard answer to all of this is to never drop packets -- ever, and then
things
> run pretty smoothly.
>
> The thing that keeps ip going, is that converse, seems to finally have the
last bugs
> killed in it, and there are actually quite a few people on converse pretty
> much 24 hours a day -- from what I've seen. Or at least if you want to
chat
> ham radio stuff on keyboards, it's better than IRC or any of the internet
chat rooms.
> (I'm kind of a jammer magnet, so I tend to stay off of repeaters.)
>
>
>
> "Rob" <NoEmail@NoWay.com> wrote in message
news:g6U_3.193228$5r2.446122@tor-nn1.netcom.ca...
> > It sounds like IP/TCP is dying and most hams are turning back to NETROM
(or
> > KNET etc)
> >
> > Rob
> >
> > "Rob Janssen" <nomail@pe1chl.demon.nl> wrote in message
> > news:slrn83nj06.ue2.nomail@linux.pe1chl.ampr.org...
> > > Cathryn Mataga <cathryn@junglevision.com> wrote:
> > > >I believe there's an advantage to running NOS over internet based IP
> > stacks
> > > >like with Linux and WIndows, that you can specify that the retry
timers
> > backoff
> > > >linearly, rather than exponentially.
> > >
> > > That is actually a disadvantage. It causes congestion collapse, and
> > > makes outside observers judge TCP/IP as a bad protocol that jams
> > > everything else.
> > >
> > > >Apparently, this is considered extremely
> > > >evil from the internet perspective, but it's a fast kluge to get ip
to
> > work
> > > >a little better when the connection is dropping packets due to noisy
> > links--
> > > >rather than congestion. (There's a linux Kernel hack, I thought
someone
> > posted
> > > >to do this, though I haven't seen anything like this for Windows.)
> > >
> > > Actually, when your links drop packets you should not attempt to
> > retransmit
> > > them from the endpoints, but you should adjust your links
(individually)
> > > to use a virtual circuit rather than a datagram technique to send the
> > > packets over that hop.
> > >
> > > Most NET and NOS versions support this.
> > > My version of NET even allows you to setup automatic selection of VC
or
> > > Datagram mode based on the independently measured packet loss rate on
the
> > > connection, separately for each direction.
> > > This works very well, without requiring inefficient retransmissions
from
> > > the endpoints that yield TCP/IP a bad reputation.
> > >
> > > >It's actually, just a hard situation all around, the more I've looked
To be continued in digest: hd_99_302D
Read previous mail | Read next mail
| |
