Hi Larry,

$.02 - is not much wort in Austria,
thought dolar exchange rates reached sky lately.
I have some doubts your ideas are not worth more, hm?

You wrote:
> One drawback of your
> plan would be that a local backup would always
> have to go back to the central
> server, and the restore would always have to come back from there.

I may repeat it for third time:
I/O agents would exchange control information O_N_L_Y
with ADSM server. They would N_O_T exchange backup data.

So the backup would NOT have to go the central server,
only control information like file name, filespace name,
file date would go.
NOT backing up to central server while
YES backing up under control of central server
is the very basis of the scenario I suggested.



I saw the announcement of V3, and while
this paper I saw is not exact enough, I believe
that distributed servers lack the
central administration point.
Even if there are tools to administer it centrally,
there are more servers to be administered,
thus multiplying necessary efforts.
I speak about things like defining schedules,
users, backup copy groups etc.
Do you have better/another informations?


Please note that my suggestion is not contradictory
to distributed servers approach at all.
Both approaches could co-exist nicely.

While IBM works on distributed serves,
I want distributed storage pools controlled
by single ADSM server.

Why not have distributed servers
with distributed storage pools ability?

Distributed servers may be best for large, distributed companies.
Distributed storage pools may be good for
small or medium sized distributed companies,
single servers with single storage pools are
likely to be best for small to medium sized one-site companies.

To be exact, $0.02 is today 0,26 ATS (Schiling),
worth about 1/120 pint of good beer in a pub.


Juraj Salak
sal AT keba.co DOT at