Hi2all.
We have ibm content manager integrated with tsm v6.3.4 on aix. Files ingested by users migrating to tsm after a year and written to lto tapes.
Once a week we must query ibm content maneger and restore about 100k files archived in tsm. The problem is in restore speed.
First and failed solution was restore using icm cm api, but cm knows nothing about tapes, drives etc. As a result we have unnecessary tape remounts and rewinds.
I know that tsm has command line utils and c api. I found in documentation two things:
- i can write C application that finds all files, sorts them by restoreOrderExt and saves to disk. But it is very hard to me to do that (little C knowledge plus no knowledge at all in aix platform programming).
- i can restore files using command line using filelist option. But as documentation states "The entries in the list are processed in the order they appear in the filelist. For optimal processing performance, pre-sort the filelist by file space name and path.". And this is absolutely makes no sense to me. All files have very similar names and paths ( e.g. {"/icm/rmdb/standard/"}/standard/some-random-content-manager-id) and how that helps to restore files in proper order i don't know.
Second path (or any solution with command line utils) is prefferable to me. Do somebody know the proper way to efficiently restore large amount of files in proper order without using C APi?
I also studing tsm database schema to dig some ideas about queries that can help with my problem, but i almost ripped my head of...very implicit object storage for me.
Any suggestions?
We have ibm content manager integrated with tsm v6.3.4 on aix. Files ingested by users migrating to tsm after a year and written to lto tapes.
Once a week we must query ibm content maneger and restore about 100k files archived in tsm. The problem is in restore speed.
First and failed solution was restore using icm cm api, but cm knows nothing about tapes, drives etc. As a result we have unnecessary tape remounts and rewinds.
I know that tsm has command line utils and c api. I found in documentation two things:
- i can write C application that finds all files, sorts them by restoreOrderExt and saves to disk. But it is very hard to me to do that (little C knowledge plus no knowledge at all in aix platform programming).
- i can restore files using command line using filelist option. But as documentation states "The entries in the list are processed in the order they appear in the filelist. For optimal processing performance, pre-sort the filelist by file space name and path.". And this is absolutely makes no sense to me. All files have very similar names and paths ( e.g. {"/icm/rmdb/standard/"}/standard/some-random-content-manager-id) and how that helps to restore files in proper order i don't know.
Second path (or any solution with command line utils) is prefferable to me. Do somebody know the proper way to efficiently restore large amount of files in proper order without using C APi?
I also studing tsm database schema to dig some ideas about queries that can help with my problem, but i almost ripped my head of...very implicit object storage for me.
Any suggestions?