| |
ZL3AI > APRDIG 25.07.03 08:54l 229 Lines 10593 Bytes #999 (0) @ WW
BID : 1663-ZL3AI
Read: GUEST
Subj: TAPR Digest, Jul 22, 4/7
Path: DB0FHN<DB0RGB<OK0PPL<DB0RES<ON0AR<VK6HGR<ZL2BAU<ZL3VML
Sent: 030724/1112Z @:ZL3VML.#80.NZL.OC #:4918 [Chch-NZ] FBB7.00i $:1663-ZL3AI
From: ZL3AI@ZL3VML.#80.NZL.OC
To : APRDIG@WW
Subject: Re: aprs2.net Tier 2 Update
From: "Phil Pacier, AD6NH" <ad6nh@arrl.net>
Date: Tue, 22 Jul 2003 16:34:6
X-Message-Number: 28
On 07/22/03, "Steve Dimse <k4hg@tapr.org>" wrote:
>The tier two network was set up by and for those people who felt that the
>central hub concept and/or implementation was flawed.
This is an incorrect statement. The correct statement would be "the Tier 2
network was set up by and for those people who felt that there are too many
inconsistencies in the Core server network at present to call it anything
close to reliable. We have one server that is in teriffic shape both in
software and bandwidth, another who has too little bandwidth and runs
software that is over a year outdated, another who has bandwidth issues
with user counts over 150 and drops users when the university decides to
play with the bandwidth, and finally another one that runs completely
different software (which seems to be working much better these days, but
still does things different from the other software package). For these
reasons, we feel the core is not reliable enough to stand alone. I simply
wish you would get your statements straight before you make a comment like
that.
>I feel there is certainly nothing wrong with the concept of the central hub
>network, and the only thing wrong with the implementation is that one of the
>hubs has a connection limit below the level it ought to be.
No one on this end ever said there was anything wrong with the concept, and
you yourself have stated before that both the Tier and the Core concept
would work for APRS depending on several factors. And as far as the only
thing wrong, see comment above.
>At present there are
>4 hubs, first, second, third, and fourth.aprs.net, of which only third has a
low>connection limit.
Second has an unreliable connection limit, as its bandwidth seems to
fluctuate. With user counts over 150, he starts building up queued data
and drops users.
>For a client wishing to connect, the simplest way is to simply use the name
>"rotate.aprs.net" on port 23. This name randomly returns one of the 4 central
>servers when looked for in DNS. For those clients that support list of multiple
>servers, I recommend putting all 4 in the connection list (1, 2, 3, and
>4.aprs.net can be used rather than first, second, third, and fourth.aprs.net if
>you prefer to save a few keystrokes!)
Now we're talking flaw, in my humble amateur opinion. Watch out if you
connect to port 23! The bandwidth you'll use on your home internet
connection to handle the full stream of APRS data is enormous! This is one
of the big advantages of Tier 2 - reduced bandwidth requirements on the
server and client ends.
>My biggest objection to the second tier concept is it increases the maximum
>number of hubs the packet must travel though between two users, to 4 from 2.
>This added complexity serves no useful purpose I can see, it only adds more
>points of failure to the APRS Internet System. If the problem is simply one of
>inadequate capacity (which is presently not the problem, third seems to be
down,>but there are still more than 200 open slots), I see the answer as adding
>additional central hubs (or moving the existing ones to sites where they can
>support higher connection limits) rather than adding an entire additional
layer.
Now, I'll just call this your opinion, and wont go so far as to say that
the Core was setup by and for people who think the Tier 2 concept is
flawed. I stand by the need for a Tiered network, particularly in the
unreliable state the core is in presently.
73
Phillip Pacier - AD6NH
socal.aprs2.net
http://www.aprsca.net
----------------------------------------------------------------------
Subject: RE: Running aprs in the car with a laptop
From: "Ben Miller" <benm@applink.net>
Date: Tue, 22 Jul 2003 16:38:06 -0500
X-Message-Number: 29
Interesting no one has mentioned noise using inverters. I'm using one to
run my laptop and get a lot of noise on my Yaesu FT-847 (mostly on HF
frequencies). I'm going to switch to a DC-DC converter for that reason.
73,
Ben Miller
W5MBM
----------------------------------------------------------------------
Subject: Re: Radio question
From: Roger Grady <k9opo@comteck.com>
Date: Tue, 22 Jul 2003 17:20:05 -0500
X-Message-Number: 30
I like my IC-2350 very much. I don't use it for digital (have done SSTV a
few times, does that count?). I think $xxx is a good price if it's in good
shape. If you don't buy it, can I have his name?
Roger Grady K9OPO
----------------------------------------------------------------------
Subject: Re: aprs2.net Tier 2 Update
From: Steve Dimse <k4hg@tapr.org>
Date: Tue, 22 Jul 2003 18:23:53 -0400
X-Message-Number: 31
On 7/22/03 at 4:34 PM Phil Pacier, AD6NH <ad6nh@arrl.net> sent:
>On 07/22/03, "Steve Dimse <k4hg@tapr.org>" wrote:
>>The tier two network was set up by and for those people who felt that the
>>central hub concept and/or implementation was flawed.
>
>This is an incorrect statement. The correct statement would be "the Tier 2
>network was set up by and for those people who felt that there are too many
>inconsistencies in the Core server network at present to call it anything
>close to reliable. We have one server that is in teriffic shape both in
>software and bandwidth, another who has too little bandwidth and runs
>software that is over a year outdated, another who has bandwidth issues
>with user counts over 150 and drops users when the university decides to
>play with the bandwidth, and finally another one that runs completely
>different software (which seems to be working much better these days, but
>still does things different from the other software package). For these
>reasons, we feel the core is not reliable enough to stand alone. I simply
>wish you would get your statements straight before you make a comment like
>that.
The only true statement above applies to third, which I have often
advocated be replaced, it was the reason I added fourth, and I still would
like to have another big hub come in to replace third, or see third move,
as it only supports 60 connections, far too few to be useful in my mind.
>>I feel there is certainly nothing wrong with the concept of the central hub
>>network, and the only thing wrong with the implementation is that one of the
>>hubs has a connection limit below the level it ought to be.
>
>No one on this end ever said there was anything wrong with the concept, and
>you yourself have stated before that both the Tier and the Core concept
>would work for APRS depending on several factors. And as far as the only
>thing wrong, see comment above.
Actually, several advocates for tier two do say that the concept is wrong,
that nothing should be connected to the central hubs except tier two
servers.
>>At present there are
>>4hubs, first, second, third, and fourth.aprs.net, of which only third has a
>low
>>connection limit.
>
>Second has an unreliable connection limit, as its bandwidth seems to
>fluctuate. With user counts over 150, he starts building up queued data
>and drops users.
That is certainly not born out in the stream statistics page.
http://www.findu.com/stream/
This page connects to the central servers (and a local copy of javAPRServ
which itself is connected to all 4 servers). Every hour, it obtains 5
minutes of data, and checks each feed against each others, looking for
dropped packets.
I watch this page constantly, and get an email warning whenever a site is
dropping more than 0.2% packets (some packet loss is expected, due to
packets appearing just before or just after the collection of the data...a
packet may appear a couple seconds sooner on one server than the others).
If there was a system-wide problem like you describe, this system would
certainly have detected it. It has not!
>>For a client wishing to connect, the simplest way is to simply use the name
>>"rotate.aprs.net" on port 23. This name randomly returns one of the 4 central
>>servers when looked for in DNS. For those clients that support list of
>multiple
>>servers, I recommend putting all 4 in the connection list (1, 2, 3, and
>>4.aprs.net can be used rather than first, second, third, and fourth.aprs.net
>if
>>you prefer to save a few keystrokes!)
>
>Now we're talking flaw, in my humble amateur opinion. Watch out if you
>connect to port 23! The bandwidth you'll use on your home internet
>connection to handle the full stream of APRS data is enormous! This is one
>of the big advantages of Tier 2 - reduced bandwidth requirements on the
>server and client ends.
No flaw. MOST people want to see all the data on the APRS Internet System.
A central tenet of Bob's APRS design, which I carried over to the Internet
System, was that the user should always be able to "read the mail" in the
finest amateur tradition. Nothing should be hidden or made inaccessible to
the user.
javAPRSFilter is availale for those who have bandwidth limitations, but
most hams still want to look at the full stream. The full stream, as bulky
as it is, is under 10 kbps, a third of a 28k dial-up line. The FUD that the
stream bandwidth is too much to handle is untrue.
>>My biggest objection to the second tier concept is it increases the maximum
>>number of hubs the packet must travel though between two users, to 4 from 2.
>>This added complexity serves no useful purpose I can see, it only adds more
>>points of failure to the APRS Internet System. If the problem is simply one of
>>inadequate capacity (which is presently not the problem, third seems to be
down,
>>but there are still more than 200 open slots), I see the answer as adding
>>additional central hubs (or moving the existing ones to sites where they can
>>support higher connection limits) rather than adding an entire additional
layer.
>>
>Now, I'll just call this your opinion, and wont go so far as to say that
>the Core was setup by and for people who think the Tier 2 concept is
>flawed. I stand by the need for a Tiered network, particularly in the
>unreliable state the core is in presently.
The numbers I present here are verifiable, the monitoring automatic, anyone
wishing is free to view this at any time. Unlike your vague accusations of
unreliability, I'm presenting facts.
And, just for the record, the APRS IS Core was set up by people 7 years ago,
many years before there were people advocting a Tier 2 network.
Steve K4HG
----------------------------------------------------------------------
Read previous mail | Read next mail
| |
