[aprssig] New Project: WHERE-IS

[Kenneth Finnegan]

Some more thoughts...

On Sun, Mar 12, 2017 at 3:41 PM, K4FHK <k4fhk at knobbe.us> wrote:

> The service has been changed today to use WHERE as the primary call
> sign. I say primary since it will also accept messages on WHERE-IS,
> WHERE-15, and still K4FHK-WI.


If you're planning on using WHERE, why also support WHERE-IS and WHERE-15?
I wouldn't really say it's been around long enough that you've got a large
legacy user base on those SSIDs. I'd suggest just using WHERE and K4FHK-WI.
Users would probably always prefer WHERE to WHERE-IS anyways, but it's up
to you to support all four aliases and explain why in the future.

Note that each result to wildcard queries will be handled with
> individual APRS messages. So if someone has 15 active SSID's and you
> query "where CALL-*", you will get 15 individual messages back.


One thing that is often overlooked for -IS attached services which spawn
multiple packets is that this won't work well for many RF users, for two
reasons:
1. Many newer APRS modems are not tested against multiple packets
back-to-back, so it isn't unusual to miss every other packet if you belch
out a long string of them at once.
2. At least Aprx, if not other RF-gate/digipeater software packages,
enforce Quality of Service (QoS) rules to protect the network from
individual stations going bonkers. This means that your 15 packets will
drain the token bucket in Aprx for the interface, and the last ten of them
will be dropped instead of RF-gated.

The lack of any sort of queuing theory applied to APRS means that it is, on
the RF side, very sensitive to traffic bursts, and not really possible to
apply any sort of fair queuing methodologies. I've caught flak before for
Aprx just enforcing a packet token bucket, but I stand behind Aprx dropping
the 6th+ packet over an -IS node having the ability to tie up the local RF
channel for an arbitrary length of time when it dumps a novel onto the
APRS-IS in message form.

I suspect you'd see a better RF experience if you metered out packets at
something like 3-5 second intervals from the source to allow the network to
move each packet and the receiver to process each one sequentially.

I do like the idea of WHERE, so please don't take me raising things you
should look for as anything but helpful.
--
Kenneth Finnegan, W6KWF
http://blog.thelifeofkenneth.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tapr.org/pipermail/aprssig_lists.tapr.org/attachments/20170312/9afe3ce2/attachment.html>