Skip to main content

Hi, we have a customer where we want to implement VMware continuous replications for his large VMs (11TB+).
We have tested it with a small VM and it was running fine. but we have noticed that during the first sync it creates a VMware snapshot to backup/create the VM on the destination site. After the VM is created and the baseline is there the snapshot is removed and continuous replication is running.
We have calculated how much time the first initialization has taken. The conclusion is that when we do this for the large VMs they will have a snapshots for almost two weeks. 
Now we are afraid that when we start a replication for this large VM and it will have a snapshot for two weeks, that after the first initialization the consolidation of the snapshot will give problems to the source VM.
Is there anybody having a solution for the VMware continuous replication for large VMs? Is there a workaround for this like seeding or something?

Regards Dominic

Hey @Dominic,

This is a tricky one - you have worked out that a baseline backup needs to occur and tracking of changes starts from that initial first snapshot. 

Does this VM have a high change rate? that will determine how fast the snapshot grows in size during the baseline, and the amassed amount of changes that have to be replaced on snap consolidation at the end.

Its an interesting question if we could seed something outside of the process to minimize the snapshot length. I’ll see if I can find somebody knowledgeable on that. 

Apparently there is a feature/change in FR26 that allows you to do the initial baseline sync without any snapshots. Adding @Pavan Bedadala to confirm!

Hi @Damian Andre 

thanks for you reply.

On the moment I'm already in contact with Swarnalatha Puligorla, he tries to setup a meeting with Pavan Bedadala and myself. He also already mentioned that the backupless feature from FR25+ should be helpful in this situation.

That is great enhancement I would say. Why is not mentioned as part of the FR25/FR26 release notes? 

Apart from eliminating the need for a snapshot based backup for the initial sync, there is another feature coming up in 11.28 that should greatly speed up the replication process. We have added support for compression, so continuous replication can compress the data before replicating to DR site. This saves space on the RP store cache, also reduces data transferred on the network between source and destination. Stay tuned. 

Nice! That sounds like a great improvement! 

Hi @Damian Andre 

thanks for you reply.

On the moment I'm already in contact with Swarnalatha Puligorla, he tries to setup a meeting with Pavan Bedadala and myself. He also already mentioned that the backupless feature from FR25+ should be helpful in this situation.

 

Would you mind please point in the right direction for this “backupless” feature?  We are using periodic replication, based off of backups, currently, and have been having trouble getting large VMs replicated as well.  

@JohnCom assuming you’re referring to IntelliSnap, what vendor are you using?

https://documentation.commvault.com/11.24/expert/37177_intellisnap_support_for_live_vm_recovery.html

Want to be sure we have a solid base before we begin.

I’ll spawn this into its own thread once we get some discussion going.

Ah ok, I am aware of IntelliSnap, but we are not using storage array based replication for this for a couple reasons, but maybe I’ll look into it again.  We are trying to keep all the VM guests replicating with Commvault LiveSync for now.  

For now, it is probably best to open a support case regarding any Livesync issues you are having.  Sounds like Intellisnap is not an option (at least for now).

Terms & ConditionsCookie settings