[aprssig] Gating Objects from Internet to RF (fina?l)

[Tom Hayward]

> One more thing. This is more than 2000 lines of data with more than 100
> chars per line, and I heard a 10 minute cycle advocated by Bob. This is a
> huge amount to dump on the APRS Internet System. You must dribble this in
> over the full 10 minute cycle, even that may be too fast. It will add more
> than 3 packets every second, but dumping more than a few lines at a time
> will really have an adverse affect on the entire system. Every full stream
> will grow by 1 GB a month, a core server can expect to see its bandwith grow
> by nearly a terabyte a month.

I'm having trouble following your math. If each full stream grows by
1GB per month, and each core server only offered full streams (which
isn't the case--most of the clients connected to core servers use
filtered streams), the max the bandwidth could grow is 300GB (1GB *
300 connections). As I count now, there are about 35 full stream
clients connected to first.aprs.net, meaning an estimated additional
35 GB per month. This is a normal-case estimate, not a worst-case
estimate.

I think before we halt the growth of APRS, we should observe the
actual impact and make changes where needed. If the core servers can't
continue to offer 300 max connections, maybe this needs to be reduced.
This wouldn't be a problem--there's plenty of capacity available in
Tier2 for IGATEs to connect to.

Trickle these objects into the APRS-IS. Any overall capacity issues
can be mitigated as they are encountered.

Tom KD7LXL