| |
ZL3AI > APRDIG 12.09.03 10:17l 256 Lines 11395 Bytes #999 (0) @ WW
BID : 1897-ZL3AI
Read: GUEST
Subj: TAPR Digest, Sep 09, 4/8
Path: DB0FHN<DB0THA<DB0ERF<DB0FBB<DB0GOS<ON0AR<ON0AR<ZL2BAU<ZL3VML
Sent: 030911/1414Z @:ZL3VML.#80.NZL.OC #:8140 [Chch-NZ] FBB7.00i $:1897-ZL3AI
From: ZL3AI@ZL3VML.#80.NZL.OC
To : APRDIG@WW
Subject: Re: Homeport
From: "Scott Miller" <scott@opentrac.org>
Date: Tue, 9 Sep 2003 10:58:59 -0700
X-Message-Number: 20
They SHOULD be permanent. They're a network service. Frankly, I think the
APRS authors ought to rip out the IGATE code from ALL of the desktop apps.
I know some of them run for months with no problems, but really... no one
else runs network infrastructure as a desktop application. Of course, at
the moment I think that leaves only aprsd. I've seen no other IGATE
services.
My IGATE setup runs on a rackmount Pentium III running Linux. Uptime is
currently 76 days, and that's only because that's the last time I upgraded
the kernel. The server, radio, TNC, switch, and DSL bridge all run off a
big rackmount UPS - no generator yet, but so far the UPS has lasted longer
than the power outages.
I think the IGATE capability in desktop software just encourages users to
run part-time gateways, which is not what we need. If I had the time, I'd
adapt my APRS-IS-to-Oracle service to be an AGWPE IGATE, but I've got too
many projects going already.
Scott
----------------------------------------------------------------------
Subject: 'How-to' on setting up an Igate
From: "Rochte, Robert" <rrochte@gpacademy.org>
Date: Tue, 9 Sep 2003 14:07:03 -0400
X-Message-Number: 21
Does anyone have a "How-to" type document available that describes how to
set up an Igate? I would like to set one up here at the Academy using a
Linux-based server, but I'm really not sure where to begin (software, etc.)
- any help would be much appreciated!
Thanks!
73,
Robert
KC8UCH
--
Robert Rochte
Director of Information Technology
The Grosse Pointe Academy
171 Lake Shore Road
Grosse Pointe Farms, MI 48236
Tel. +1 313-886-1221 x155
FAX +1 313-886-1418
www.gpacademy.org
----------------------------------------------------------------------
Subject: Re: Homeport
From: David VanHorn <dvanhorn@cedar.net>
Date: Tue, 09 Sep 2003 13:36:58 -0500
X-Message-Number: 22
>I think the IGATE capability in desktop software just encourages users to
>run part-time gateways, which is not what we need. If I had the time, I'd
>adapt my APRS-IS-to-Oracle service to be an AGWPE IGATE, but I've got too
>many projects going already.
i see your point..
i run ui-view as an igate, and it's uptime has been pretty good.
it's also my 'b' system where i run my ham/hobby apps, separate from my
main desktop, where i earn a living.
as such, ui on 98 meets my needs.
the system has a modest ups for the computer, which has carried me through
all power outages in the last 6 years, plus dc backup for the radio and
tnc, which also runs my other radios.
what we're having to face now, is that the network is maturing.
before, anyone that would run a digi or igate was welcome, because we had
more holes than network.
now, the network in most places is pretty full, and we suffer from too many
digis, and too many igates. i have a standing offer to my local
coordinator to take down my digi or igate, if they are ever a problem, or
no longer serving a need. i put them up because we were in a 'black hole'
here, and the nearest digi couldn't reliably cover the area. if some other
station can do a better job, then i'll gladly take my gear down. no
problem. i would hope that all the network ops would feel pretty much the
same way.
----------------------------------------------------------------------
Subject: Re: Homeport
From: Bob Bruninga <bruninga@usna.edu>
Date: Tue, 9 Sep 2003 14:56:47 -0400 (EDT)
X-Message-Number: 23
>"Doomed to failure"? Come on, Bob. Usage of a home port is much less
>likely to cause problems than the incessant use of RELAY,WIDE7-7,
>TRACE7-7.
We agree here completely. What's doomed to failure is if we let users
anywhere flood IGates of their choice anywhere else by simply picking a
spot on the map *without and understanding* of what they are doing.
>Knowledge of the local APRS network should be focused on utilization,
>not topology for the everyday user.
I didnt say that. I said for the person who wants to go on distant
travels and wants to flood his packets back to his IGate (his own
backyard), then, -he- should know and understand his local network. He
represents maybe only a few percent of the "everyday user"...
>... Digis change. Requiring each user to learn the ever changing
>topology to make use of APRS somewhat also defeats one of the features
>of APRS, doesn't it. In other words, what you are saying is "I don't
>have to know what resources are nearby to use them in APRS EXCEPT if I
>want to use a home port."
Yes, that is what I am saying. If one is going to FLOOD an IGate with
his traffic from a-far, I want him to KNOW what he is doing...
>Finally, lets get down to the "brass tacks" of what a homeport as you
>propose it is good for. It is so ALL of my packets will be seen on RF
>in my home location when I am not at home, right? Messaging is already
>handled transparently for the user so that doesn't factor in. Hmm. Is
>that really using APRS as a tactical protocol?
For the distant traveler who has reason to be in real-time tactical
communications with his wife or friends back home, I would say yes. ANd
it is better than if he was trying to do it with WIDE7-7...
>At least Wes' subscription idea would be initiated by a ham in the local
>area to do some tracking of the "remote" station. Seems to me this
>would be much less likely to cause unwanted QRM (at least someone in the
>local area is stating "I want to see that station's packets" :-).
I disagree here also. Lets say that there are 50 users in my local area.
And most of them has a friend somehwere else in the world that they think,
"wow, thats neat, I can subscribe to your positions and you can subscribe
to mine and we can see each other across the country while we chat on
IRLP. Now I have 50 more stations pushing data into my backyard that I
dont need or want on our local network.
>And what about the occasional "I want to know where X is?" Then send a
>message to X and if they are in range of an IGate, you will know where
>they are.
We already have that. If I send out a Query for the position of WB4APR-9,
doesnt any RF IGate that hears it respond with the position? I thought
they did. APRSdos does.. (but APRSdos doesnt see the APRS-IS so that is
a moot point)... But I assumed all other IGate software did.
Thus, I see no need for opening up all kinds of processes for TARGETing
Igates for being flooded from afar, other than in the case of HOMEPORT
which is a request by a local for packets back to his own area, which I
contend is a far less complex and dangerous issue.
Bob, WB4APR
----------------------------------------------------------------------
Subject: Re: 'How-to' on setting up an Igate
From: Adi Linden <adil@adis.on.ca>
Date: Tue, 9 Sep 2003 14:32:58 -0500 (CDT)
X-Message-Number: 24
>Does anyone have a "How-to" type document available that describes how to
>set up an Igate? I would like to set one up here at the Academy using a
>Linux-based server, but I'm really not sure where to begin (software, etc.)
>-any help would be much appreciated!
Have a look at aprsd: http://www.wa4dsy.net/aprs/index.html It's what I
am currently using. It has good documentaion, IMHO.
There is also http://sourceforge.net/projects/aprsd/ which is aprsd with
ax.25 support. Haven't used that yet.
Adi
--
Adi Linden | va3adi | va3adi@adis.ca | http://adis.ca
+-----------------------------------------------------------------+
IRLP Node 2590 in Sioux Lookout, Ontario
----------------------------------------------------------------------
Subject: Re: Homeport
From: "Wes Johnston" <wes@johnston.net>
Date: Tue, 9 Sep 2003 15:49:00 -0400
X-Message-Number: 25
----- Original Message -----
From: "Bob Bruninga" <bruninga@usna.edu>
>We agree here completely. What's doomed to failure is if we let users
>anywhere flood IGates of their choice anywhere else by simply picking a
>spot on the map *without and understanding* of what they are doing.
I think these PUSH (IGATE designator in your path or homeport) / PULL
(subscriptions) ideas are too good to not implement. What we need to do
here is to focus on the ways in which the system can be abused and come up
with controls to limit the damage by malicious or niave operators. Just
think, if we pull this off, 90% of aprs users will be using a path of just
RELAY. I know James Jefferson did that study on how many hops everyone was
from an IGATE. If we do this IGATE thing correctly, our on air hops will
plummet and we'll have smaller and smaller "cells" of localized aprs
activity. I'd like to see the network one day where as I travel to/from
home to another city, you'd only see me on the air a)in my hometown and
b)within about 20 miles of my present location. I'd also like to see this
done with me having to contact an IGATE operator and asking him to gate all
my packets out (kd4htw in South Carolina did this for me when I was in
Philadelphia a few years ago).
Yes, I understand there is room for screwups and people can knowingly or
unwittingly flood a local RF network... BUT by and large, most people will
run shorter paths and direct their traffic to people who actually care about
that traffic vs. broadcasting in a hugemongous circle hoping that their
intended audience gets the packet. MIC-e directional routing was supposed
to do this, but never caught on.... now we have the opportunity to direct
our traffic to a specific town instead of just "north of me" or "4 hops
around me in all directions."
This is where intelligent IGATES come into play... Some igates already have
throttling in effect. Packets could be throttled on a station by station
basis. If k4lid sends a packet every 5 seconds, the IGATE in the area he is
flooding will limit him to something more reasonable like 1 packet every 3
minutes. The other 35 packets he sent are just dropped on the floor at the
IGATE. Have a hystersis... He can transmit up to once a minute, but if he
exceeds that, he's limited to once every 3 minutes - just an idea. Don't we
limit PING packets on a network in a similar fashion to prevent DoS attacks?
As much as I hate to say it, there is also the possibility an IGATE owner
would get tired of seeing k4lid tunneling out of his IGATE all the time and
simply budlist the guy.
Perhaps too, there is the need for a QoS feedback? The IGATE could send a
message back to the sender stating that the network he was reaching was
nearing saturation and ask that he reduce his rate.... like the path police
idea from a few weeks ago.
Also, the possibility of denied service comes to mind. The german FlexNet
system won't let you connect if your TXD is too long for example. I realize
that one day there may be a legimate need for some of our more stupid paths,
but really, if APRS is (in time of emergency) a local thing, then a min path
of RELAY, WIDE does the trick. Stupid human tricks like WIDE7-7 and WIDE,
RELAY need to just have their packets dropped on the floor by the likes of
digined... not modified or corrected... DROPPED. When the niave user
figures out that he's not in the network at all because he did something
stupid he'll fix it.
Wes
----------------------------------------------------------------------
Read previous mail | Read next mail
| |
