I'm not sure how we could do it without a major paradigm change.
Xylogics did it because the intelligence was server side - the ACP/ERPCD
system did most of the 'thinking', the Annexen themselves were fairly
'dumb' in this regard. ACP would pick a box to do the dialout, etc.
But Livingston's paradigm has always been to have an intelligent NAS with
little reliance on a single server when at all possible. The dialout
system on a Livingston is to establish an interface, so there is a route
in the table that can be broadcast, but to keep the interface in a suspended
state. To the LAN there *is* a connection - there is no difference at all
between a point where the connection is really there and when it is waiting
to dial. When traffic comes in for a suspended interface, then the PM
makes the dial and routes the traffic.
Having multiple boxes claim the route would be messy...
And if one box that claims the route is full, how could it tell another
one to make the dial? And if it can, will it forward the traffic? Will
it tell the other unit to start taking the route and drop it itself? That
works ok with OSPF, but it could wreck a RIP network...
Hmm...
-MZ
-- Livingston Enterprises - Chair, Department of Interstitial Affairs Phone: 800-458-9966 510-737-2100 FAX: 510-737-2110 megazone@livingston.com For support requests: support@livingston.com <http://www.livingston.com/> Snail mail: 4464 Willow Road, Pleasanton, CA 94588