[aprssig] Why Not "Gate in Vicinity"

[Greg Dolkas]

I agree with this.  The earlier post that talked about the IS as "just a transport" may have been accurate early on, but things have changed.  

In corporate networks, the network equipment is far more than just a dumb transport.  Quality of Service processing can prioritze traffic based on both priority and available capacity.  Defensive measurs are also employed to keep rogues from taking down the network.  These features are increasingly important as network load increases, and as new services are deployed and depended on by their customers.

With our severely limted and shared wireless channel bandwidth, and the wide scope of the IS, I expect that traffic bandwidth shaping will be necessary to maintain the stablity of the APRS network overall.  

So, what traffic rules do we need, and where shoud they be implemented?  How about Greg' thoughts, below?

[another] Greg  KO6TH

Sent from my trusty iPAQ.

-----Original Message-----
From: "Gregg Wonderly" <gregg at wonderly.org>
To: "TAPR APRS Mailing List" <aprssig at tapr.org>
Sent: 12/25/11 6:19 AM
Subject: Re: [aprssig] Why Not "Gate in Vicinity"


On Dec 23, 2011, at 4:48 PM, Lynn W. Deffenbaugh (Mr) wrote:
> The only, IMHO, truly valid and "safe" fix for too-fast beacons is to fix the clients that are issuing them.  The network isn't responsible, but the -IS to RF IGate operators are, so the software they are using needs to provide suitable assistance.

I think that there are more important concepts that can help.  I think that there should be some bandwidth analysis used in the gate to RF processing.  In particular doing bandwidth used calculations, over a particular amount of time for all gated traffic, as well as for each station, would help it perform some simple throttling techniques.

For example, gated traffic might be limited to 30% of all bandwidth available in a 5 or 10 minute period.  If more traffic occurs, then just like a Level 3 network path, it is just dropped, summarily.   

Also, for individual call signs (without SSID), perhaps a maximum of 10% of the bandwidth could be gated.

This seems like some simple book keeping, and the counting can be dynamically managed to not be a large consumer of memory.

Gregg Wonderly
_______________________________________________
aprssig mailing list
aprssig at tapr.org
https://www.tapr.org/cgi-bin/mailman/listinfo/aprssig