"scanner -i" will take too long for DDBoost & DD-AFTD devices

I recently wanted to compare scanner operations when using the same command for different NW device types. During my tests I found out that "scanner -i" will perform extremely slow for a backup device if it resides on a Data Domain (DDBoost & DD-AFTD devices).

I verified this for NW 19.3.0 up to the latest version 19.5.0.2 - it might well be that this behavior is the same for a bunch of earlier NW versions.

It took a while to verify and detect - along with Dell/EMC support - the reason for this performance degredation. As it turns out, the 'scanner -i' process actually issues a kind of 'ping' for each discovered file. Consequently the overall performance will decrease especially if a save set/volume contains millions of small files which is nothing unusual today. Just to give you an idea: scanning one volume containing 'only' 10mio small files may take several days for a DDBoost device ... for the same save set on a DD-based AFTD the performance is even slower

 

Another issue you should keep in mind:

At the end of the disaster recovery routine nsrdr the programs asks you whether you want to scan your volumes right now. If yes (or if you confirm the default setting) nsrdr will silently execute a 'scanner -i' in the background which of course will end in the same effect as described above. So whenever you use nsrdr you ...

  -  should avoid the built-in scanner process

  -  run 'scanner -m' on the volumes to prevent the loss of the save set information

  -  but ran 'scanner -i' on new save sets only if this is necessary.

 

This issue has been addressed and escalated - it will be solved soon although a specific date is not available yet. Until then, it is a good idea to avoid 'scanner -i' and the automatic 'scanner' process at the end of 'nsrdr' for the DD-bases device types.