About tsm for file backup

[joluinfante]

Hi!
I have to migrate to TSM my old backup method.
Now, I need to support a zero level of files (and incremental, next days), on different computers, in a combination of different flavors of Linux and Windows.
I did complete the tsm4ve implementation (to backup the machines in esx environment), but now I do backup others machines, are not running in a esx environment. I know I can do this with tsmc (baclient) program.
* There is some mechanism where centrally sched indicate which files in which machines have to backup? or should I run the tsmc on each machine?
* Have the case of an Oracle database, where, for the moment, I can not use an agent database. then my backup is to stop the database, and make a backup of each of the datafiles (which are basically many large files, of which only change a few blocks). I understood that there was a mechanism "backup only changed blocks in a backup file type". There is that?

Thanks a lot for your responses

 

[moon-buddy]

* There is some mechanism where centrally sched indicate which files in which machines have to backup? or should I run the tsmc on each machine?

TSM is not an agentless backup system and for that matter, you need to install a client on each node (server). The files that needs backup can be listed on the dsm.opt or dsm.sys for Windows and *nix environments, respectively.

* Have the case of an Oracle database, where, for the moment, I can not use an agent database. then my backup is to stop the database, and make a backup of each of the datafiles (which are basically many large files, of which only change a few blocks). I understood that there was a mechanism "backup only changed blocks in a backup file type". There is that?

Change block tracking is not part of the TDP for Oracle legacy backup. More so, if you cannot use TDP, your choice is to dump the database to a flat file and backup the flat file.

 

[marclant]

First and foremost, I strongly recommend you take some TSM training if you never administered TSM and want to be successful at it.

* There is some mechanism where centrally sched indicate which files in which machines have to backup? or should I run the tsmc on each machine?

Yes: https://www-01.ibm.com/support/know...com.ibm.itsm.client.doc/c_sched_overview.html

* Have the case of an Oracle database, where, for the moment, I can not use an agent database. then my backup is to stop the database, and make a backup of each of the datafiles (which are basically many large files, of which only change a few blocks). I understood that there was a mechanism "backup only changed blocks in a backup file type". There is that?

This is true, it's called adaptive file backup. However, it doesn't work well with databases because files have to be smaller than 2GB and the delta is usually too large and the whole file needs to be backed up anyway. So you will have to backup all the data files. Alternatively, you could use RMAN to do an online backup of the DB to a file and then use TSM to backup the file RMAN created.