[plug] RAID-1 vs rsync vs ? for high availability

[Denis Brown]

Dear PLUG...

I'd like to make a high availability Linux box to handle (among other
things) groupware - calendars, etc.   I've located and will work through
the High Availability Linux HOWTO but I'd also, in parallel, like to
garner some feedback from members who may already be using alternative
solutions to a cluster.

It seems to me that the primary cause of failure in a modern PC is going
to be mechanical failure - HDDs, fans, etc.  Can monitor fans for failure,
activate a blower to compensate, etc.   But on the HDD scene something
more sophistocated seems called for.   RAID-1 looks promising but maybe
there are better solutions.   Some parameters:

High availability - MTTR 30 minutes or so in the case of HDD failure.
Data integrity - must survive a HDD crash.
One box initially - may be able to add boxes later using CODA or
clustering.
Two HDDs of same size, both IDE one per bus, no other devices on IDE.

The RAID HOWTO points me in the direction of RAID-1 and by using
persistent superblocks would mean that I can boot off the array.
Assuming "sensible" failure conditions RAID-1 looks to have good data
integrity chances.

Life would be simpler if I used plain vanilla HDD technology -- dispensed
with RAID and went for a primary HDD and a secondary, one per bus, and
synch'ed the two using something like rsync perhaps as part of a cron job
(big ?? here, not yet read enough.)

The primary application is a groupware server and while it would be nice 
to have redundant boxes, etc that might not be possible in the early days
of the project -- I'm wanting a toe-in-the-door to strut some Linux stuff.
I have omitted the obvious aspects such as choosing a high quality mobo,
adequate psu, not overclocking the cpu, running the show from a ups...

Thoughts would be appreciated :-)
Cheers,
Denis