We are looking at ADDSM and have several questions.


(1) Will Andrew  File  System support  be provided  for  the HP  backup/
    archive client?  The announcement  letter refers  to the  Andrew File
    System only for  the AIX/6000 backup/archive client.

(2) Does the  Andrew File System support include  incremental backups? Is
    volume location and mount point recorded?

(3) What are the plans for supporting the OSF Distributed File System?

(4) Can someone  elaborate  on how  ADDSM works  with  file servers?  The
    following question From Rick Chang's  survey would suggest that there
    are some changes that are needed.

> 11. Would you like a client to be able to restore files that were backed up
>     (to an ADSM sever) by a file server?
>
>      __ 1.A must __ 2.Very helpful __ 3.Nice to have

    Can  the root  user  restore any  object to  any  location? Is  there
    security provided to prevent one  user from restoring another's files
    (unless the restoring id has read permission)? Are timestamp and file
    permissions preserved when an object is restored? ACL lists?

    Is there any  support for system or group administrators.  If not can
    this  be handled  with an  exit. For  example one  could have  an MIS
    person in charge of accounts payable.  You would want them to be able
    to  act as  a surrogate  for the  end users  of the  accounts payable
    software. Or an instructor over a set of students.

    In an environment where workstations  are not personal and people can
    log onto one of n machines, how does one handle backups and restores.
    assume  the  end-user  file  directories  are  stored  on  a  set  of
    file servers. Do  you have a  separate backup for each  user directory
    tree on the server. Or is the server backed up as a separate unit. If
    the server  is backed up  as a separate  unit can the  end-user still
    request restores of her/his file(s)?


(5) Performance and capacity planning:

    I looked at  the red book on ADDSM  but it only gives a  small set of
    samples  for performance  or  capacity planning.  Is  there a  better
    source of information for this.

    Does  anyone have  an estimate  for the  average space  used for  the
    catalog space for  each object. I believe someone  reported 600 bytes
    per file.  But this seems  high for an  average when the  objects are
    compressed.

    For 1 million files x 600 bytes x 2 for duplexing x 10 for
    10 generations or cycles this is 1.2 x 10 to the 10th power.

    The red book only talks of a MVS server. Are the performance numbers
    any different for a VM server.

    The  red book  shows numbers  for  backing up  one workstation.  What
    happens when one backs up a large  number of workstations and
    file  servers  at  once?  Can  the  server  work  on  more  than  one
    workstation at once? Is  the VM server limited to the  use of one cpu
    at a time? Greg Piney reported that it takes about 10 mins to back up
    a 1  gig machine.  If one  assumes that  one could  only work  on one
    backup at a time, this would mean about 16 hours for 100 workstations
    each with 1 gig. What happens with 1000 workstations of 100 meg each?
    What happens if there is 200 gig for 1000 workstations and
    file servers?

(6) Futures

    Can  anyone talk  of the  differences between  the new  OS/2 and  AIX
    servers  and  the VM/MVS  servers.  Will  there  be new  features  or
    functions? Or will there be less?

    If one has more than one server will each one be completely separate
    from the others. Or can the end-user see them as one virtual unit?
    That is if one had 3 vm servers or 10 aix servers can a client
    do a search and restore operation against all the servers with
    one operation.


                          Thanks in advance for any help that you can give

                          Len Boyle