Quantcast
Channel: VMware Communities: Message List - vSphere™ Storage
Viewing all articles
Browse latest Browse all 5826

Re: Bizarre growth of disk usage when moving thin provisioned VMs between Datastores

$
0
0

Found the answer which is as simple as: compression of the zfs-folder on the storage server.

 

I had (falsely) assumed storage-side compression/deduplication would be transparent from a vsphere perspective.

-> Question: is that the "normal" behavior also on other systems that support comp/dedup (somebody over at spiceworks-forum where i also posted this issue said storage-side space savings on his NetApp-Storage through comp/dedup weren't visible from vsphere).

 

Another reason for me to rule out compression was the huge differences in the sizes. Overall compression ratio on the Storage is 1,37 - compression ratio for the example-VM is 4,98. -> this brings me to the next (non vmware) question:

 

Used space on the virtual disk as displayed inside the VM is 2,1GB - the raw virtual disk file is 11,98 GB (uncompressed) -> free space (that did hold data and got deleted at some point) is roughly 9,8GB. These blocks do not contain zeros (since zeroing+punchzero on the vmdk eliminates the space almost completely) - still the compression ratio on this data is apparently insanly high. I'm wondering what kind of data is on the disk. Data is mostly static on this VM with only a tiny mysql-db that is generating IO.

 

-> Are there any tools that allow deeper analysis of freespace on an ext4-partition (like Zero/Non-Zero ratio)? Are there any tricks that could help in keeping the virtual disk files smaller without manually zeroing inside the VM and running vmkfstools on the vmdks?


Viewing all articles
Browse latest Browse all 5826

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>