request for info

[Bdale Garbee <bdale@col.hp.com>]

> We are currently trying to avoid NETROM/ROSE wars by proposing a 5 year
> comprehensive plan which will take advantage of new technology while not
> disrupting the current LAN users.  We have the following items on our
> wish list:

>       1. high[er] speed backbone.  9600 baud as a minimum.  maximum=?

Kantronics recently sent me some preliminary sales poop on their DVR4-10 70cm
data radios.  Price near $400 list.  Something better than 10w out, a better
receiver than the 2-2, and designed to allow operation up to at least 19200.
They will have a "modem" card for the DataEngine that basically wave-shapes
the datastream to/from the radio.  Interesting, since it'll give you off the
shelf 19200.  Hope it fairs better than the AEA boxes on 220Mhz did...

That means that hardware will exist (plus or minus interfacing to your
favorite switch widgets) for 1200-19200 off the shelf.  I assume you know about
the WA4DSY 56kb modems, which I think are pretty slick, but which I see as a
local channel modem more than a link modem (though opinions differ widely on
the right answers for link hardware for different uses).

Here in CO, we're working on a plan that would put 10Ghz links in place across
the state.  Exact data rate not determined yet.  Exact link hardware still
being debated.  But based on the Grace Communications packet switches, with
1200 and 9600 baud ports at each hop along the way for user access.  In at
least one area (mine), there will also be 1200 baud full duplex repeater
capability on 2m with a port into the switch, and 56kb cross-band full duplex
repeater capability between 220 and 430Mhz (input on 430).... all as local
user channels.

>       2. TCP/IP support for as many sites as possible.

The model we like is to build a fully IP backbone, and provide AX.25 user
services at each node, not all that differently in some respects than the
NET/Wrong and TheNet model, where an AX.25 user connects to his local node,
then has some set of things he can do.  In many cases, what is really happening
is a TCP connection is being opened on the backbone to a service providing
machine elsewhere, or to another node... but the user needn't know or care
about that at all.

>       3. redundant paths for trunks.

Absolutely!  Wish we could do that first pass, but we can't.  Too little
population density for the distances involved... may be able to use 9600 baud
ports as link ports when the 10Ghz stuff is rained or snowed out... but...

>       4. easy access for AX.25 users.

See above.  And see my previous messages to the list about the Grace
hardware.  The standalone boards, particularly with some changes reportedly
in the works, seem like the exact right answer for building switch sites that
need to survive some number of upgrade cycles in the future before needing to
be replaced... and they're juicy enough to support a fair amount of protocol
and route processing...

Bdale