[nos-bbs] A.X25 then, A.IP now ??? - thinking outloud (fwd)

[Andre v Schayk]

maiko at pcs.mb.ca schreef:

>A bit of history first, to give you all an idea of where I'm leading ...
>
>During the 90's, I designed, developed, and installed automated banking
>machine and point of sale networks, using a software *switch* to route
>transactions to the appropriate authorizing institutions, etc.
>
>Many of the ATMs (ABMs) used the X25 protocol to communicate to the
>switch. The whole concept of X25 was really cool (still is) and I got
>to really liking it, becoming quite experienced in setting it up, and
>writing/porting comm drivers for a variety of different Unix flavors.
>
>That probably explains why packet radio (A.X25) grew on me.
>
>At the tail end of the 90's, our company saw a new breed of ATMs (ABMs)
>coming out in the market. In particular, NCR (NDC), was starting to ship
>units that gave us the option of using TCP or UDP over IP, instead of
>the usual X25 protocol. This was a turning point, and for many of our
>customers that could afford it, X25 was on the way out, IP was the
>new way of doing it.
>
>Warranted, alot of new customers could not afford to purchase new
>ATMs (ABMs), so X25 was still in use here and there, but then again,
>alot of the used atms only had SDLC or BISYNC to offer, making X25
>less popular anyways (from my experience).
>
>Here's the main point I want to bring up !
>
> In the amateur radio field, A.X25 (packet radio) was created from
> the X25 specification. I'm surprised that no one has come up with
> an A.IP (ip over radio) spec to parallel the industry movement from
> the X25 protocol to the IP protocol. Or has someone brought it up ?
>
>When things settle down, and winter starts, one of my projects is
>to perhaps implement an A.IP protocol (actually nothing more than
>raw IP with a SOURCE CALL and maybe ONE DIGI field). The nature of
>IP really only requires us to id ourselves I would think. That should
>make it legal, right ?
>
>Implementing it would be very easy. I should think we could continue
>to use EXISTING tncs, running either KISS or SMACK firmware, to make
>this work. KISS just delimits the data that the tnc is putting out or
>getting in, right ?
>
>Of course A.IP would have to be on a dedicated non A.X25 frequency,
>unless there was agreement between the A.X25 spec and A.IP to have
>some type of flag at the very first byte of data.
>
>Anyways, just some thoughts. It would be neat and alot more usefull
>with the modern days apps out there, to have literally raw IP going
>out and coming in a TNC. I think so anyways ...
>
>I'd like to have comments on this, what do you all think ?
>
>Regards,
>
>Maiko Langelaar / VE4KLM
>  
>
I might be mistaken but isn´t IP a level 3 routing protocoll? if that is 
the case you would either need a simple level 2 protocoll to let it ride 
on, (slimmed down ax25 perhaps) or have all listening stations pass it 
directly to de ip stack weather it´s needed or not.
 From your discription it sounds like a combination of them, a slimmed 
down ax25 just to id and having all stations pass it to the IP stack 
(there is no recipient).
Passing it directly to the IP stack has a downside and an upside, the 
upside is that all posible routes are tried without the need to tweak 
the routing tables, just select the interface in the routing table (arp 
has no meaning anyway), the downside is that all posible routes are used 
all the time including congested routes.

My feeling is that in the end it might not be worth the trouble and just 
use ax25 datagram, this already eliminates a lot of the ax25 overhead.

73 de Andre PE1RDW