[ax25-layer2] Re: The BASIS of NSR is questionable

[Robert Bruninga]

>>> aprs at kd4rdb.com 09/28/05 2:19 PM >>>
> it seems that your philosophy is to encourage
>everyone to run reasonable paths but leave 
>enough bandwidth left over that you can run 
>a longer path when needed. 

Yep, exactly.  Just like the fundamental rule in
HAM Radio, use the minimum power necessary
to communicate.  But be prepared for the
ability to communicate as needed.

>NSR brings that to a screeching halt....
>the problem comes in when 'everyone' wants
>to play - and "everyone" does want to play.... 
>all the time.  Or when 'everyone' sets and 
>forgets their paths that work in one part of the
>country and not another.

I think NSR is trying to fix the same problem we
FINALLY got around to fix with the New-N paradigm.
That is, for the first 12 years of APRS, there was
NO discipline, NO mechanism for SYSOP management
and NO single path recommendation.  It was a free-
for-all with all kinds of legacies, and evolved paths. 
Finally, inl Nov 2004, I had just had  enough!  It was 
time to DO something.

The New-N did several things (and is working well):
1) It greatly simplified paths to one path WIDEn-N
2) It gave local SYSOPS abuse-trapping capabilities
3) it ELIMINATED the biggest load on the system by
    a factor of 2 to 4 (dupes caused by RELAY & WIDE)
4) It alerted all APRS users to the new simplified
    system so that education works.

and the combination of 1,2,3&4  has had striking
results.  Also, it turns out the biggest LOAD problem 
on the network was NOT long-hop-out-of-area
packets but the huge load due to the RELAY-WIDE
dupes.   

NSR arose about the same time as a way
to eliminate the out-of-area load.  But it turns out,
that is no longer a significant problem.  This
is because with the (1) simplification to one
universal path recommendation, (2) the ability
for SYSOPS to trap long hops and abuse, now 
then (4) education of users WORKS.

>NSR fixes the whole "what path do I use in 
>this area" question.  

So does New-N.  Use WIDEn-N.  And look at the
BEACON of the last DIGI you heard and see what
N is receommended in this area if one is so clueless
to not understand that a BIG METRO area is 2-2.

>Now, [under NSR] if you want to go further than 
>that area that a sysop has defined..., you must
>use a gateway.  

In my mind using 3 hops to get to my daughter in
the next state with a directed path of DIGI1,DIGI2
DIGI3, is far more efficient than ANY OTHER system
including NSR for the simple reason, that NSR remember
is ONE-PAT-FITS-ALL, so therefore it will FLOOD the
starting LAN just to get the packet to a gateway
and then gate it to the NEXT Adjacent LAN and
FLOOD it to get to the target.   My original 3 hops, 
3 packets becomes 16 packets under NSR (8 local 
packets and 8 packets in the other LAN.  This 
violates the rule of using the minimum "power"
to communicate.

>In terms of internet, this means that you can 
>talk to everyone in your "workgroup"/"subnet" 
>easily, but to go outside those bounds requires 
>routing thru a switch.

It also means that every packet no matter how
closely it is intended will FLOOD your local "group"
and will FLOOD the other end of the gateway too.
I do not see this as a net gain.

>But please don't think that NSR stops your packet 
>at the edge of your local area... It seems that 
>the big black unknown here is the gateway.
>For ease, it can be an igate. We already have 
>100's of people with igates and cable modems... 

Yes, but somehow it hardly ever works for me!
I find the reliability of anything through a double
gateway to be extemely fragile..  It simply
does not work well and is so iffy as to be not worth
doing.  Planning an NSR system that is not only
totally dependent on the whims of the local SYSOP
but then also dependent on PC's and the whims
of local IGates is the kind of experience I have
had in ham radio that tells me that that is not the
way to go in general to go 3 hops.

>but the gateway can also be an intelligent
>network switch that can divert only the necessary 
>traffic to a higher bandwidth (or lesser used) 
>segment of the network.  There is no reason
>this gateway can't be a 1200 or 9600 baud link 
>to the next town on 6m, 70cm, etc.

Sure, there is no reason, other than it just wont
happen.  You are dreaming.  I dont know about
you, but in general, anything in ham radio that
depends on that kind of massive infrastrucute
just simply wont happen.  It s a great idea, a
great way to design things and lots of fun to drean
about.  But it just wont happen.

Just look at your remaining email... Yes, there
have been all kinds of solutions and ideas, but
it just doesnt happen....

>a year or two ago I expanded on your lan/wan 
>idea and came up with a scheme that would allow 
>digipeaters to share 4 full duplex frequencies.  
>It would require each site have 2 receivers and one
>transmitter.  See slides #8 thru #20 in
>http://www.kd4rdb.com/kd4rdb/Backbone%20routing.ppt .  
>It even works if some of the nodes cease working.  
>With an idea such as this in place, we simple need 
>the level 3 software to take advantage of the hardware.

>In so far as a given NSR network not being friendly 
>to other digi's that are put up when the main ones 
>are down due to a storm,... you program the
>replacement with it's callsign. 

And in the mean time, under NSR nothing works...

But with New-N, every digi is a digi that supports
WIDEn-N and so the digi and the hardware are
transparent.  

>It seems that on the one hand, you (Bob) want 
>to call APRS a local resource, and on the other
>hand, you want to be able to check in to a net 
>in Philly which is beyond_your_ local area around 
>Annapolis.  

Yep, late on a thursday night, I think that is 
something that I should be able to do, using a
directed path of 3 hops that is FAR MORE 
EFFECIENT than flooding my local LAN with
traffic I want to go north.

>the 40 stations aloha circle gets you 21 miles 
>around Annapolis.  How do we define local?  
>Can we be fair and say that local is the same 
>area for all of us in a given town?   No fair..

Yep, which is EXACTLY the problem of NSR.
A local SYSOP gets to define it only to suit
his personal view of who he thinks should
be allowed to communicate where.

In multiple overlapping metro areas, like
DC, Baltimore, Philly, NY to Boston, there
are no sensible boundaries.  Only the user
knows where he wants to communicate.
There are people that live in Baltimore
that work in DC and vice versa.  Hence the
definition of a fixed bounary LAN is
useless.  In APRS the definition is "the local
area CENTERED ON YOU"  And in APRS,
"your needs" are what drive yoru path.

>the demise of packet was not necessarily the
>internet.  Indirectly, it was all the darned forsale 
>ALLUS messages that flooded the BBS systems...

You got that right!

Bob

Robert Bruninga wrote:
>>>>jewen at shaw.ca 09/27/05 8:56 PM >>>
>>
>>What if the network were able to determine 
>>for itself what the "ok list" is, by measuring 
>>the current network load? 
> 
> 
> That  would completely undermine the objectives of
> NSR to give stability to the network.  A network is
> useless if it is always changing and even worse,
> if no one can tell what it is doing.  The key is consistency.
> 
> There have been lots of ideas proposed for APRS to 
> have the digis simply cut the flow (ignore some packets)
> to maintain only the flow that the network can support.
> 
> But that is a ludicrous way to run a communications
> network for its own sake.  Yes, it works well for 
> people who think that APRS is just a means for
> filling their maps up with ICONS.  And they dont
> care who's or how far.  Just so long as they get
> a nice flow.
> 
> But it is useless for the end user that needs to
> be able to look at the network and see where
> he needs his packets to go. 
> 
> 
> Bob
> 
> 
> 
> _______________________________________________
> ax25-layer2 mailing list
> ax25-layer2 at lists.tapr.org 
> https://www.tapr.org/cgi-bin/mailman/listinfo/ax25-layer2 
> 
> 
> 
> 

_______________________________________________
ax25-layer2 mailing list
ax25-layer2 at lists.tapr.org 
https://www.tapr.org/cgi-bin/mailman/listinfo/ax25-layer2