Ted Spendlove wrote:
>
> But, lets stop and think about this for a minute.  We are postulating a THREE 
> level disaster
> here.  One, we lose the client.  Two, we lose the server. Three we have a bad 
> copy pool tape.
> As long as my clients and my server are not in the same physical location I 
> cannot justify the
> extra processing to create two offsite copies for my data center.

Our ADSM server is an SP2 node. Many of the clients are other nodes in the
same SP2, and the rest are in the same room. For a number of reasons there
is no realistic prospect of changing this. At my institution, the political
difficulties of obtaining space for a remote ADSM server would be enormous,
and the cost would be substantial if the political difficulties could be
overcome. My management would be very reluctant to give up the performance
advantages of sending backups over the HPS (the fast, short range network
used to connect nodes within an SP2). It would be expensive to provide
either a tape robot or operator coverage for a remote ADSM server.

With the ADSM server and clients in the same room, I am only postulating
a two level disaster, and even that statement is not as reassuring as it
sounds at first hearing. A copy pool tape could go bad months or even
years before a fire or similar disaster led to an attempt to read the
tape, and I would probably have no idea that the tape problem had
occured until and unless the computer facility disaster actually
happened.