Julie, do you truely mean a new DISK volume ?

     If so I hope they don't want to keep very many copies!

     Say you grant them 7 isolated disk copies...
     Have the client give you the size of the directory.
     On 7 different disks, slice out a storage pool volume that big.
     Define 7 different storage pools (probably want to name them something
     like SPECIAL_MON, SPECIAL_TUE, SPECIAL_WED, ...)
     Use an archive job with the -ARCHMC=xxxx to pull in the data.
     Set an administrative command schedule to rotate through the storage
     pools.  Now to avoid having to activate the management class daily you
     could define a storage pool by the name SPECIAL_POINTER and assign NO
     volume(s) to it, then on Monday at 00:01 run something like "upd stg
     special_pointer next=special_mon" to actually point the data over to
     the isolated disk volume used in the definition of the special_mon
     storage pool.  Do the same sort of activity each day of the week.
     If the archives have a 7 day retention, each pool would be cleared by
     expiration on the day it was to be reused, you would have to make sure
     expiration ran prior to the current day's archive.
     Now the special management class you would have to define... I would
     not even put a backup copy group in it... having it there would
     probably just screw things up at some time or another.  And that
     management class's archive copy group is where you would use
     DESTINATION=SPECIAL_POINTER.

     Now this SHOULD all work 'cause when the archive starts to roll in and
     adsm tries to preallocate storage pool space it will see that NONE
     exists in SPECIAL_POINTER and roll to the defined NEXTPOOL to obtain
     the space (which is being rolled through the SPECIAL_MON, SPECIAL_TUE,
     and so on...)

     I have something kind-o-like this but select groups of TAPES per day
     was good enough.  Here I just sent their archives to a single, special
     diskpool.  Once the archives were done I migrated the data to a
     special tape pool.  Once the migration completes I do :
     update volume * acc=reado whereacc=readw wherestgpool=pr3590
     wherestatus=filling,full
     So each day the migration starts with a fresh scratch tape (this is about a
     10+ tape activity) the last, partially filled tape is kept from being used
     the next day by the command mentioned above.  And once the data on the
     tapes expires, the tape rolls back scratch thus resetting the "readonly"
     condition.  I also generate a list of these volumes daily so I can tell you
     what tapes belong to any given day's recovery (retrieve) set for this data.

     Hope this helps...



______________________________ Reply Separator _________________________________
Subject: how to fall backup one directory of one machine to a disk
Author:  a9500672 (a9500672 AT SP1.MACARTHUR.UWS.EDU DOT AU) at unix,mime
Date:    8/23/98 5:43 PM


I have a adsm server and normally we backup all the node incremently in
two steps ( first backup to the disk and magrate to tape. ). For the some
reason, I be asked to backup fully one directory of a RS6000 into a new
disk volume every day.

How do I do it??

Any comment will be appreciated

Thanks in advance

Julie