VMware vSphere

 View Only
  • 1.  What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 11:09 AM

    According to this article : VMware Knowledge Base

    A snapshot should not be used for more than 72 hours.

    This seems quite specific. What happens at 72 hours and after than makes having a snapshot so bad?



  • 2.  RE: What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 11:41 AM

    I hope continuum​ can answer this, he has an older post I can't find, that explained this well. For me its a general time frame that doesn't let most snapshots get to big, but we also watch the size and delete any if they get too big before that.



  • 3.  RE: What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 11:44 AM

    Read this, its someone asking the same questions, and look at the last comment

    Re: Snapshot Persistence



  • 4.  RE: What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 02:00 PM

    There's nothing specific that happens after 72 hours, it's merely a guideline or rough suggestion to a snapshot's usable lifespan. As long as the snapshot exists, it continues to grow, increases I/O latency, and can endanger other VMs on the same datastore. Generally after 72 hours (or before) it should either be committed or reverted.



  • 5.  RE: What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 02:41 PM

    When it is stated that a snapshot continues to grow, is that indefinite or is there a maximum size?

    My thinking is that it should be unnecessary for a snapshot to grow much beyond the size of the original vDisk. I.e. For a 100GB vDisk for example, once the snapshot has recorded changes to every block in the original vDisk, it shouldn't need to grow any larger, and would therefore contain 100GB of deltas, plus any overheads for associated metadata and logging.



  • 6.  RE: What happens with snapshots after 72 hours...

    Posted Feb 06, 2019 02:54 PM

    The snapshot can grow up to the size of the disk upon which it's based. So, a VM with a single 100GB vmdk can have it's snapshot take up to 100GB consuming 200GB on the datastore (not factoring in other overhead). So you can get into an out-of-space condition on your datastore rather quickly. This is one reasons snapshots should never be left unattended.