We have an NS702G connected to a CX600 with approx 13 CIFS File Systems.
One file system (FS) is 1.5TB in size and 98% full. The FS was created from the "clar_r5_performance" storage pool.
Although we could easily extend this FS, we do not want to - this is because in the event of a DR, the time to restore (via NDMP backups) would be too long (1Hr per 100GB).
Q: Can you please tell me whether running a file system at close to capacity impacts the end-user performance?
Additionally, we are upgrading to DART 5.5 (from 5.4) this weekend and I was told it is bad practice to upgrade the NAS with a FS close to full capacity.
Q: Rather than just telling the users it's bad practice, is there anything more concrete that I can tell them to get old data archived?
Our filesystems are almost always at 98% and there is no impact. If they get to 100%, then you will have troubles just like any other filesystem. Users will not be able to write data until space is available.
As far as upgrading to 5.5. The pre-upgrade script will complain that you have filesystems at 98%, but it is not a show stopper. If the Celerra root volume is over 80% (not exactly sure on that number), they cannot do the upgrade since there is not enough room for backups and the new code.
Has your question been answered? If so, please click on Helpful or Answered at the top of the messages which have assisted you. This will award points to the person who posted that message: 6 points for Helpful and 10 points for Correct You can mark only one message as Correct and up to two as Helpful.
There is not a significant performance impact to running filesystems at 98% full, however, there are possible issues that can result from this if the filesystems ever needs maintenance, for instance (fsck, or in windows terms, chkdsk) ran on it. Because there is not sufficient room to place files that have been lost in the inode tables, they will not be able to be placed in the lost+found directory, and will just be lost. There is also supplementary information about the filesystems stored with the filesystems. This area is very important to CIFS since it holds key data on the security placed on the files/directories. As you create new files, entries are made in this area. If there is not sufficient space to put this, the file security information may not be set to what you want it to be.
It is best practice to maintain 15% free space in a filesystem at all times. Your backups will only take as much time as there is data in the filesystem, regardless of how large the filesystems is.
It is also best practice to keep filesystems under 500GB in size. This is because of backup/restore times, and the fact that if filesystem maintenance is ever needed on the filesystem, it can be done in a reasonable time frame. Currently, this can be done at a rate of about 300GB/hour.
sscss0
15 Posts
551
1
Posted November 10th, 2006 09:00
As far as upgrading to 5.5. The pre-upgrade script will complain that you have filesystems at 98%, but it is not a show stopper. If the Celerra root volume is over 80% (not exactly sure on that number), they cannot do the upgrade since there is not enough room for backups and the new code.
Hope this helps.