@Zevgeny
For a few snapshots it's fine - the issue comes when you have large numbers. Most of my first-hand experience with this specific issue comes from Hyper-V but the logic applies here.
This is from VMware but the concept is the same and has some good reading on snapshots in general.
Running a virtual machine on a snapshot for extended periods of time can cause instability and data loss.
https://kb.vmware.com/s/article/1015180
The underlying issue stems from the nature of snapshots. With none you have a single VHD containing all of the data for the VM, once you take a snapshot a new delta VHD is created with all new data written going forward from the time of the snapshot. So, in order to have a complete VM you have the original VHD + the delta VHD.
This is fine in smaller numbers, but once you start having 10, 15, 20+ snapshots, every one of those represents an additional VHD in the chain. This results in performance degradation because you're reading from multiple potential files as only new data is written to the deltas.
This is actually from XO's blog: https://xen-orchestra.com/blog/xenserver-coalesce-detection-in-xen-orchestra/
