From: Adam <""4thwormcastfromthemolehill\"{at}the field.near
the bridge">

Geo. wrote:
> "Mike N."  wrote in message
> news:phcss211l88e3ahd3r4r0gf991ll3l4fn7{at}4ax.com...
>
>>   In the case where you can afford to take it down long enough for an
>> image backup, it's OK.  If it cannot be down 30-60 minutes for the
>> backup,
>> this is unacceptable.
>
> I'm having trouble imagining that someone would use pc class hardware
> for a server where you can't schedule an hour of downtime. I mean it
> seems to me that kind of service would require a different class of
> hardware, possibly done with PC's and round robin or something of the
> like but a single PC?
>
> Geo.


Also depends on being able to move services selectively & basically
using routing.

E.g. you have server it has a

DB & An appserver on it serving lots of users so you need to be able to
smoothly & easily moving the appserver over to a new machine & then
point it back at the old db & then reroute users to the new appserver
(which is still using the old db).

Then transfer the db & rebuild the db on the second machine

Then switch the app server to point to the new address for it's data
&........now you're good to go 


Adam

--- BBBS/NT v4.01 Flag-5