Dennis writes:

> There would be lots of benefits gained by backing up remotely like
>  that.  You could have scripts written to on the half hour determine
>  which files have been modified and ship them to the backup machine.

Exactly so. What's important to remember in all of this is how few files
actually get modified in a given time period, thus the amount of information
needed to be transferred is likely to be relatively small, thus these
repetitive "delta backups" are likely to be very quick.


>  Is the performance for FTP much faster then DSCOPY, or did you just not
>  buy the NS3000iX license?

We own DSCOPY, but we've never used it. As a developer and not a "true" user,
it has always been our general approach to always develop around those
products and services that are native to the HP3000 and never require the
user to have to purchase anything else, if it's at all possible to do so.

My original interest in FTP grew out of other concerns, QCTerm particularly.
I actually never thought about using DSCOPY internally, so I can't say how
the two compare. If my thought process had been reversed, seeking out a
machine-to-machine backup strategy for only our own use, I probably would
have considered DSCOPY first.

In either event, using DSCOPY or FTP, the general process would be the same.



>  The only gotcha that I could see is that if all the nodes/houses are
>  hardwired using ethernet would be fears of lightning or electrical
>  problems in one house cause problems for the other boxes.  If the LAN is
>  implemented using fiber pairs and media convertors no problem....

The LAN that runs between the houses and the garage (which we actually call
the "server room") is a standard coaxial thin-LAN, sheathed in a plastic
conduit, buried at least 18 inches in the ground at its shallowest, but more
normally 24 inches. Hubs with BNC connectors tap into the thin-LAN at various
locations. The thin-LAN is, of course, terminated with 50-ohm terminators at
each end.

Because the coxial cable represents its own electrical universe, and its
impedance is so low, and because it is buried below the surface, an
electrical strike right on the ground above it should induce virtually no
transient current. And because the only taps into the thin-LAN are made with
hubs, should a significant amount of current arrive at one of the hubs on one
of the RJ-45 arms, I would expect the hub to be a perfectly good sacrificial
lamb and pass little or none of the transient into the thin-LAN.

Lightning travels odd paths, but it is essentially always seeking the best
ground possible, following the path(s) of least resistance. In this case, the
thin-LAN is "floating" at a relatively high impedance off of the true
electrical ground, with a very low internal impedance. That's a little bit
like asking water to flow uphill, through a thin pipe, only to be met by a
dam. We shouldn't expect to see a lot of current flow.

Wirt Atmar