We are using Tivoli Storage Manager for Backup, especially TDP for Oracle / RMAN for Online Database Backups and it is working fine.

The problem is the $BSE filesystem, because there are million of files from different companies and VRCs, objects of DDIC and so on.

The customer wants a daily full backup from $BSE, although less 2 % are changed and this means a ressource problem of backup database, because 1 file
is 1 entry.

From my point of view there are 2 possible solutions.

1. Image Backup of logical volume from $BSE, but at the moment baan appl. is stopped, processes are killed which accesses the Filesystem and then lv backup will create 1 big file on tape. I am looking for a solution that can make a snapshot without unmounting filesystem with backup / OS tools.

2. Backup incremental with tsm client, means only files that are changed since last backup will be written to tape. I can set a parameter that file will be backed up, although a process handles a file, but this normally no problem for files , only for database.

Any suggestions or ideas ?
Is the a baan session ttadv... to stop temporarily Development/Upgrade activities , so i got a consistent backup of ddic-objects ?

thx in advance

Hi there,

you can do a tar or cpio of the $BSE directory and only save this file on tape.
Then in case something happens, you have to do one step (unpack) more, but I don't think that you will often meet the request of restoring a single file into $BSE.

hth

Norbert

Hello,

thank you for reply.

Your suggestion to compress $BSE in one File is similar to a logical volume backup.

But you have to know that because there are million of files in thousend subdirectories it takes hours of time to make this tar-file.

How can I be sure that i get a consistent backup of this ?

An example: If an developer change some important things in DDIC, or create a VRC, then the files be changed or created in this time , when the process is running, I think

Hi mainframe1,

admittedly our $BSE may be a lot smaller than the one in your environment, but the cpio here took 90 seconds for 70.000 files and the archive is 1,5 GB big, if zipped it has 400 MB.
Assuming that your developers don't work all night, it should be possible to create such an archive before starting the actual backup on tape.

-----start
cd $BSE (might not be set !!)
find . -name "*" -print | cpio -ovB 2> /dev/null > PATH/NAMEOFARCHIVE
-----thats it

... change 2> /dev/null to 2> PATH/LISTOFCONTENTS if you like

So if things go astray, you can reload the file from tape, unpack the archive and only have one entry in your tapeware database

hth

Norbert