[nos-bbs] Re: AXUDP and APRS

[Andre v Schayk]

Bill Vodall schreef:

>>But you see here is where things are philosophically different between
>>the APRS world and the Conventional packet world. APRS allows for blind
>>exchanges of data and links between systems, where as the conventional
>>packet systems tend to caution on the side of security (ie, AX*P tunnels
>>are pre-arranged and there are IP address checks in place, etc). Clients
>>do not negotiate for an AX*P tunnel out of the blue.
>>    
>>
>
>Ah - that's the point.  APRS is growing and has a substantial following.
>Conventional packet is about dead.  Winlink is growing and it is very
>similar to APRS with a simple RF or web based authentication system.
>The additional caution and security is not needed and is severly
>hindering the use of the system.
>
>I've given up on the ampr.org system.  Far too complicated and outdated.
>For all that disagree -- send me an Email from an AMPR.ORG system on
>RF...   It used to be straight forward.  Today it'll take a week to make
>it happen IF YOU'RE LUCKY!
>
>The AX*P system is a good mix.  It's packet - which has the "fun factor"
>and interfaces well to the real world (AKA The Internet) and to amateur
>RF.  It's a good way for folks to contribute resources from their broadband
>home setup with the local and regional communities.
>
>  
>
>>BUT wait, are we getting ahead of ourselves ?
>>
>>What would we use this new AX*P system for ?
>>    
>>
>
>Communications.  Experimentation.  Chat.   Everything and anything.
>
>  
>
>>What's the plan Bill ?
>>    
>>
>
>I have a local "plan" and an idea of how it can be universally leveraged.
>
>Locally, I'm active with what's left of the packet and TCP/IP communities.  We're
>making a bit of RF noise with packets and a few experimental systems are being
>worked on.
>
>On the other hand I'm also talking a lot with the Section Manager
>and Section Emergency Coordinator about what hams are doing.  It always
>seems we could be doing so much more if we made better use of digital
>technologies.
>
>Putting that all together has given me a personal goal of building
>a small network of Linux servers for the Section with
>a TCP/IP "backbone" and RF ports to local neighborhoods.   My idea is to
>use Netrom (to start with) as a "User Friendly" resource identification technology.
>Anybody logging in to these servers and bringing up a NODES list would see
>all the other servers and resources for the Section.
>
>That's why I'm interested in getting AXUDP working to provide a transport for
>the Netrom (or AX25) packets between the servers.
>
>My goal is a small clean "working" network.  Just local and neighboring region
>nodes -- that all work.   Not the flood of thousands that usually don't work
>from the existing AX*P system.
>
>How's *NOS fit in here?   The backbone servers would be basic Linux
>systems with AX25.  *NOS has some unique capabilities that makes it
>useful in the layer above the core servers.  I'd imagine some of the
>core servers would also be running *NOS and other user clients that
>didn't want the battle of a full Linux box would make use of it.
>
>One of the keys to make this successful is that it has to be easier then
>it is now.   Any techniques we develop towards that end will help similar
>networks elsewhere.
>
>  
>
>
>We do have some incredible technology here.  If we can make the initiation
>fee a bit lower it'll be a good thing...
>
The same level of securety as used in the aprs-is is good enough for the 
lawmakers so I guess it is good enough for us too.

I have been thinking a bit about this the last few days and I feel a 2 
tier network would work fine (a set of hubs and leaves and clients 
connecting to the leaves)
The hubs might need a lot more bandwith then the aprs system especialy 
when people start sending large files. to make sure no loops accure no 
packets should be send back to the link it came from and there has to be 
only one outgoing link active at the same time.
to reduce the bandwith needed for the leaves only the packets that need 
to go to the clients of the leaves need to be send to the leaves.
For securety and setting up the routing I think a control link shoudl be 
set up in a simular way it is used in ftp where the login and commands 
are done on one port and the data is send over another flexable link. 
the setup system could be an initial route is setup to the call a client 
logs in with and aditional calls clients want to get can be requested 
trough the control link, leaves can pass the needed calls on to the hubs 
in the same way as leave are technicaly clients to the hubs, the hubs 
don´t need to know what is behind the leaves that way.
the only systems that get all the trafic is the central hub.
the system can be something like:

clients->leave->hub->hub<-hub<-leave<-clients

when a leave connects to another leave that seccond leave becomes 
automaticly a hub, it´s a bit like how an irc network works only the 
system does not have to know who is connected, all it needs to know is 
what needs to be send to what incomming link and to send everything 
except what came from the outgoing link to the outgoing link.

if there are any questions or comments about what I proposed let me know 
and I´ll try to clearify it.

73 de Andre PE1RDW