This post is more than 5 years old
3 Posts
1
2116
How to calculate snaphot sizes
When you create a snapshot, my understanding is it does not actually copy the data and instead just makes a copy of the references to the data. As data is updated, the original data is left unchanged and the new data is written to a new location (which essentially creates a copy). Adding new data to the parent volume should not affect the existing snapshot.
Assuming that's true (and not too over-simplified), how much space does a snapshot actually consume?
Example:
does it immediately reserve the full volume space?
does it only consume space if/when data is updated on the parent volume?
is there some minimum overhead space requried for each snapshot?
I'm looking at creating snapshots on a schdule as part of a backup plan and need to understand how much space the snapshots will consume.
PowerFlex Mark
40 Posts
0
August 21st, 2017 01:00
Hi,
ScaleIO snapshot uses redirect-on-write, which means after a snapshot is created, any new writes to the source volume, will be redirected to a different location, while the original location is used by the snapshot.
There is a performance penalty for first writes. ScaleIO allocates spaces in 1 MB chunks. In case of snapshots, if a first write comes in for a 1MB chunk, the whole chuck will be copied, regardless of the IO size. But subsequent writes to different sectors in the 1MB chuck do not need to be copied.
The Frontend views provides detailed information about the frontend objects in the system including snapshots.
Details about net capacity in use, and snapshot tree including volume name, and the date and time of creation can be found.
Hope this helps.