I think replication and shadowing combine to good effect.

In the primary server room, one would like to have the shadow on a
second system (NFS mounted to the first), with automatic changeover
scripts, as described by Alex.

This setup would cover against H/W failure of the primary node, of
whatever nature, be it disk, network card, CPU or whatever. Clients
with shortlived connections (web) would not notice anything, long lived
clients would have to reconnect, using identical connection settings.

Replication would cover against more major setbacks, such as a fire in
the building or a plane plowing through the server closet.

> A number of our clients work with NFS-based shadows, and statistic is
> the following. There were approximately about 9 power-failures
> analyzed during 3 years, and in 4 cases main database was broken up
> to being unrecoverable with gfix/backup/restore process. Shadow has
> always been health.

> There are two symmetric systems, and when primary server can't
> start, secondary activates it's own fbserver, and after playing a bit
> with IP addresses clients even do not notice, that they work with
> another box. [...] such pair of servers help people to continue
> their work after the failure almost immediately.

Alex,

This is an important topic for marketing. Part of the evaluation will
always be the high-availability options for Firebird. It would be good
if we would have some sort of whitepaper describing your setup and
failover scripts. It would make for good copy for the FBDeveloper
magazine.

If you can describe to me in a bit more detail what your setup is,
especially how automatic activation of the shadow is achieved, I will
try to put that in some sort of whitepaper. I'm sure others are willing
to help as well.

Paul