I once used a mainframe based backup product called 'Harbor'. This
product had the feature you mentioned were it used pointers to a single
copy of a file that existed on multiple files.

What I would ask is how does the application store all the 'other' data
associated with the file,  ie was it NTFS on server 1 but not on server
2, security keys, compressed or not etc etc. Storing this info would not
seem insurmountable but then it becomes a question of the cost of CPU
and time to process this information verus the cost of storing multiple
copies.

Kelly's point of excluding files such as WINWORD.EXE or the entire base
OS  is valid assuming you have a method to ensure that the base
environment is standard across all servers - in other words The Standard
Operating Environment.

For most companies implementing or who have implemented the SOE
structure, they have to have a method to 'push out' the SOE therefore by
default solving the first part of the BMR restore process (ie building
the base OS including the TSM client)

My opinion (for what it is worth) is that with the SAN environment
becoming common place the standard ideas for BMR processing needs to be
re-worked. The following is something I would like to test further.....

(For the following points I am referring to the EMC product offerings as
that is what we use and have had success with)

1) For critical or servers with large amounts of data they should be on
the SAN therefore a localised failure in the server hardware will
require at worst minimal data restore. If you do not boot off of the SAN
then only the OS would need to be recovered. Maybe even data that was
corrupted due to an unclean shutdown.

2) Restoring from total SAN based data corruption can be facilitated by
disk mirroring with Timefinder / Snapview type products. Major databases
are a classic candidate including the TSM server itself.

3) For full DR and a combination of the two previous scenarios the
remote mirroring (ie SRDF) is the option for those with multiple sites.
Combine this with booting off of the SAN then your BMR solution would
involve nothing more that sourcing the hardware and you SAN zoning.  The
second site option could even by supplied by you DR service provider in
the form of a dark site.

For TSM clients not on the SAN.......

Base Operating System (the SOE environment) could be on the SAN
therefore in the event of a DR:
Source you Hardware and Install a Temporary HBA card

 Connect & boot your new hardware from  the SAN.  The SAN based OS
would have the appropriate TSM  authority to restore the 'lost server'.

Restore the entire image of the lost server to the local disk on the
new hardware.

Remove the HBA & reboot

With standardising hardware and the hot pluggable drives in most new
equipment the restore server might be a permanent SAN connected host and
you could simply pull the drives and put them in your replacement
server.

Peter Griffin
Sydney Water









>>> stephen.a.cochran AT CAHIR DOT NET 02/14/02 03:46am >>>
I'm not an expert with ADSM, but does ADSM have any of the advanced
features that this article talked about? Can it back up a file only once
for all the computers that have a copy of it? Or will it back up
Winword.exe 5000 times for my whole organization?

We're starting to look at what it will take to offer backup for the
entire campus, and some of these advanced features would be desirable.

Steve Cochran
Dartmouth College


-----------------------------------------------------------
This message has been scanned by MailSweeper.
This message has been scanned by MailSweeper.
-----------------------------------------------------------
-----------------------------------------------------------
-----------------------------------------------------------
This e-mail is solely for the use of the intended recipient
This e-mail is solely for the use of the intended recipient
and may contain information which is confidential or
privileged. Unauthorised use of its contents is prohibited.
If you have received this e-mail in error, please notify
the sender immediately via e-mail and then delete the
original e-mail.
-----------------------------------------------------------
=========================================================================