[aprssig] Re: Another Source of Network Overload ?? -- SuperSUPERLAN
Robert Bruninga
bruninga at usna.edu
Fri Jan 7 14:30:34 EST 2005
>>The blanket statement was that "packet at 9600 is only twice
>>as fast as 1200."
>
>That is not true. Saying so as a general statement can lead folks to
>think it's not worth the effort to persue better systems.
It *IS* very true for the context in which it was stated. And that
was in response to Jim's post about all his many years of experinece
with adapting BBS, LAN, and NODE's to 9600 baud.
If a typical packet is 80 characters long and takes 670 ms at 1200
baud plus 300 ms of TXD for about 970 ms and an ACK is only about
20 bytes long, takes 160 ms plus 300 ms or about 460 ms, and throw
in about 100 ms for processing and turn around, then it takes about
1530 ms per packet at 1200 baud.
Simply changing the packet rate to 9600 baud on the same
radios reduces these to 421 ms for the packet and 320 ms
for the ack for a total of about 741 ms for the complete packet.
The difference, 741 ms verses 1530 ms is close enough to
"only twice as fast". I stand by my post.
We ALL KNOW that the only way to make 9600 baud
work faster is to get the TXDelay down. BY AN ORDER OF
MAGNITUDE. And this means ALL NEW RADIOS.
And again, we can cry till the cows come home that we all
need all new radios, but it just ain't gonna happen and I don't
intend to wait 10 years until it does. So far we have been
waiting since the mid 1980's and HAMS just arent buying them.
And if they didn't buy them in the mid 1990's when HAM RADIO
PACKET provided all their email before the internet came out,
they surely arent going to go buy them now...
So YES, much better 9600 baud systems can be
dreamed of and designed. But it aint gonna happen
and I'm not waiting for them, and I'm not going to waste
my time designing systems for future vapro-radios that
only 1% of the users will buy and use...
Lets move out with what we got. It works PERFECTLY
for the 9600 baud super-lans I am proposing...
de WB4APR, Bob
More information about the aprssig
mailing list