Unsolved
This post is more than 5 years old
1 Rookie
•
13 Posts
0
1286
September 9th, 2019 06:00
Multiple Snapshots and dedup for migrate volume
I have a couple of question of Compellent features:
- Are multiple snapshots cascaded - i.e if I take snapshot 1 at 1pm and then write at block A and then take snapshot 2 at 2pm and write at block B is there a COW operation to copy the "old" contents of block B into both snapshot 1 AND snapshot 2 or just snapshot1 (so that snapshot 1 only contain COW block from 1pm to 2pm and snapshot 2 contains COW block from 2pm onwards)
- If I take a snapshot and then create a volume from this snapshot and then migrate this volume (Local Copy > Migrate Volume) to a new volume, then is dedup used so that initially not much space is used or is dedup not used for this operation - I am guessing not, but I just thought I would check
No Events found!
DELL-Sam L
Moderator
•
7.6K Posts
0
September 9th, 2019 13:00
Hello Mike B Storage,
Please let us know if you have any other questions.
Mike B Storage
1 Rookie
•
13 Posts
0
September 12th, 2019 05:00
Thanks for reply,
So are you saying if I take 2 snapshots of a data volume, then when I write data to the volume it will do a COW for that block into BOTH snapshots so the "old contents" of the block is copied to both snapshots so is duplicated in the snapshots and if this is the case is dedup done here?
There is a good description of "cascading" snapshots here https://sort.veritas.com/public/documents/sf/5.0/hpux11iv3/html/vxvm_admin/ag_ch_snapshots_vm7.html#170118,(space-optimized means COW) - the upside being, it takes up less spaces and involves less writes, the downside being you have to recursively read snapshots when reading and if you delete a snapshot you have to copy (or merge) the snapshot data.
Mike B Storage
1 Rookie
•
13 Posts
0
September 18th, 2019 08:00
Below is a screen shot of a volume which has 3 daily snapshots and one manual. If snapshots are independent then each write written today copies the old block into every snapshot and so all older snapshots must be bigger than newer snapshots as older snapshot contain ALL the data in the newer snapshots plus their own blocks, else if snapshots are cascaded then each write written today copies the old block into only the newest snapshot, so for me it is clear the snapshots are cascaded as the snapshot space for the 16th Sept (7.32GB) is bigger than the older 15th snapshot (5.46GB) .
What I don't understand is why the first snapshot on the 14th is the whole size of the volume - does this mean the first snapshot is a full snapshot and not a COW snapshot, so there is 587GB used by actual volume and another 587GB used by the snapshot volume and is this why the Space Recovery column says "Yes" for this snapshot. If this is NOT the case and ALL snapshots are copy-on-write (COW), then why is the first snapshot shown differently, appearing to use all the space.