copy savesets from tape A To multiple tapes..

If you do that, let me suggest that you better switch off multplexing (parallelism=1) for the tape backups.

If you use multiplexing (assume that you have a parallelism of 4 and 4 save sets active to the same pool at that time), such is good for backups (keeping the tape in motion). However, if you clone a single save set you travel along the whole tape to read (100%) but you only need to copy 25% (a single of the 4 save sets). Consequentl, such will extend the necessary clone time.

An even better approach would be that you backup to disk before you clone to tape.

I am kind of new, i going to  research about  multplexing .

Well my approach is to copy first to a pool on datadomain  disk and then at  the end of the month () copy that saves sets to another historic pool on tapes.

Thanks

Of yourse you can clone save sets from a (DD) volume to tape. But you should do that in an unmultiplexed way - save set by save set. Scripted cloning would be an optimal way to achieve that.

'Multplexing' is the result if you send data from multiple streams to the same tape drive/volume at the same time. The result is a mix of data streams (aka interleaving). This was used when backing up to tape with the intention to keep the drive 's t r e a m i n g' if one stream is not capable to send the data fast enough.

Now, if the data already exists on (DD) disk, the data path to tape drives usually are fast enough. Multiplexing here would not make sense. It is even contra-productive if you think about restores. Why? - because it is more unlikely that you will recover all interleaved save sets/streams at the same time. But the tape drive has no choice as it works as sequential media. So it must read/travel parts of save streams which are not needed. The result is a longer restore process.

Hi, first at all sorry my rusty english. I going to try to understand:

1. So to avoid multiple streams i need to  create a cloning script  (any example?)
2. I understand than the tape drive works sequetialy, actually when i schedule the clonings job i do it like this example:
   1. Saturday cloning job takes savesets from  weekly pool_01 (on dd device)  to offsite weekly_01  pool  on tape 01
   2. Sunday cloning job takes savesets from  weekly pool_02 (on dd device)  to offsite weekly_02 on tape 02

Thats ok?

1. No - you do not necessarily need a script. Another alternative would be to set the target session for a tape device to 1. But this could be the wrong value for other operations.

    However, scripting is my personal preference and you can run a separate script for each disk device you want to clone from. That's another way to achieve parallelism.

2. For a NW Windows server, Powershell would be the ideal environment for these scripts.

    Whatever method you use, NW-wise you only need 2 core statements:

       - mminfo to create the clone save set list and save it to a file

       - nsrclone to run the clone process

   Such script would consist of these core lines (samples):

$out_file = $global:path + "ssids_for_" + $global:volume + ".txt"

$command = "mminfo -s " + $global:nwserver + " -q 'level=full ,!incomplete,!suspect,copies=1,volume=" + $global:volume + ",savetime>-1weeks"' -r 'ssid' > " + $out-file

invoke-expression $command

$command = "nsrclone -v -F -s " + $nwserver + " -J " + $snode + " -d " + $snode + " -b " + $pool + " -y " + $global:month + "/01/" + $global:year + " -S -f " + $out_file

invoke-expression $command

Hi again, thanks  for the explanation and  the scripts samples, i m going to try.