Hi. In a VMware environment, I have VSA on virtual machine. Running HotAdd backup, if the VM backing up is on the same host of VSA, the final “remove snapshot” is very fast (few seconds). If the VM backing up is on a different host of VSA, the final “remove snapshot” is very slow (some minutes) and in the meanwhile the VM is unresponsive.
Someone can explain to me why this happens?
Thanks
Best answer by Damian Andre
RubeckDK wrote:
So when doing a manual snap and removal using native VMware tools (ie GUI, PowerCLI etc) the same issue shows, or? (If doing it manual, please let it run off the snap for approx. the same time as the backup would take so the delta *00000x.vmdks will have a reasonable amount of data that needs to be merged back into the org. vmdks)
/Rubeck
Good suggestion! - Also ensure not to snapshot VM memory but enable quiesce to simulate the same type of snap a backup takes. You are 100% right that you should wait between create and remove that the typical backup operation takes to accumulate a similar type of snapshot change delta.
Sometimes it's not possible to simulate as the same load conditions could vary at different times of the day - this issue is usually pretty consistent, however. It’s a shame VMware has been unable to fix it for such a long time. Here are all the possible solutions I can think of:
Change to NFS or NBD transport
Deploy a VSA on each host (this avoiding cross-host snapshot mounts)
Linux VSA is a good option if you want to avoid windows licensing
It can also help with load balancing. Commvault will auto-select a VM proxy on the same host where possible
Use NFSv4 to mount the NFS datastores if your storage supports it
Filesystem agent in-guest, or perhaps use selectively where the stun causes operational issues
So when doing a manual snap and removal using native VMware tools (ie GUI, PowerCLI etc) the same issue shows, or? (If doing it manual, please let it run off the snap for approx. the same time as the backup would take so the delta *00000x.vmdks will have a reasonable amount of data that needs to be merged back into the org. vmdks)
So when doing a manual snap and removal using native VMware tools (ie GUI, PowerCLI etc) the same issue shows, or? (If doing it manual, please let it run off the snap for approx. the same time as the backup would take so the delta *00000x.vmdks will have a reasonable amount of data that needs to be merged back into the org. vmdks)
/Rubeck
Good suggestion! - Also ensure not to snapshot VM memory but enable quiesce to simulate the same type of snap a backup takes. You are 100% right that you should wait between create and remove that the typical backup operation takes to accumulate a similar type of snapshot change delta.
Sometimes it's not possible to simulate as the same load conditions could vary at different times of the day - this issue is usually pretty consistent, however. It’s a shame VMware has been unable to fix it for such a long time. Here are all the possible solutions I can think of:
Change to NFS or NBD transport
Deploy a VSA on each host (this avoiding cross-host snapshot mounts)
Linux VSA is a good option if you want to avoid windows licensing
It can also help with load balancing. Commvault will auto-select a VM proxy on the same host where possible
Use NFSv4 to mount the NFS datastores if your storage supports it
Filesystem agent in-guest, or perhaps use selectively where the stun causes operational issues
We use 3 different kinds of cookies. You can choose which cookies you want to accept. We need basic cookies to make this site work, therefore these are the minimum you can select. Learn more about our cookies.