Windows server 2019 Hyper-V: Replica VS Checkpoint VS Screenshots

MVP

I'm looking for a comparison between these 3 features of Hyper-V which all do the same thing and have the same purpose. Replica and Checkpoints are a lot more similar to each other. they save the state of the VM and they give us the ability to Revert back to that state in case a disaster or failure happens.

 

so why would someone use Replica while Checkpoints provide the same result in a cheaper and less complicated way? Thanks

5 Replies

Hi @HotCakeX 

 

The functions seem similar, but the use cases are different. Checkpoints capture a VM state in a point in time so you can roll back. Not that I'd recommend it, but you can keep a VM running on a checkpoint for a couple of weeks and go back to that point if you needed to.

 

Replica's are used more in a disaster recovery. Replica is on a entirely different host and is synchronized with the primary VM regularly (I think between 30 seconds - 15 minutes) so that if the host your primary VM resides on explodes, you can fail over to the replica and only lose 30 seconds - 15 minutes of data, so the recovery time between disaster striking and being back up and running is (in theory) pretty short.

 

While you can have multiple replica points you can go back to, I believe they can only go back as 24 hours; useful if a server has been hit by a cryptolocker, but not useful if you need to roll back further.

 

I tend to think of checkpoints as helping if there is a problem with the VM, whereas replicates help if there is a problem with the host.

 

Hope this helps,

 

Mark H

Hi, Thank you for your response. after reading it I have some more questions.

should cluster failover and replica be used together? because as you said replica is used when there is a problem with the host (node). if the VM is part of a cluster failover, then in case of a node/host failure, another node/host will take control of that VM. so with that in mind is it any good to use both replica and cluster failover together?
(the actual VM files reside on another server using ISCSI)

 

or maybe replica and failover can actually work together?

In my opinion cluster failover can remove point of failure in hardware (all except for the hard disk where VM VHDX is located on).

and then replica can remove the point of failure in the hard disk by keeping a copy of that VM VHDX file on another ISCSI server?

 


@HotCakeX wrote:

 

In my opinion cluster failover can remove point of failure in hardware (all except for the hard disk where VM VHDX is located on).

and then replica can remove the point of failure in the hard disk by keeping a copy of that VM VHDX file on another ISCSI server?

 


You are correct - failover clusters are useful when a single host/node goes down; providing another host can reach the shared storage the VM Config and VHDX files reside on, it can crack on working.

 

Replicas are typically done between two data centres - The primary is in Datacentre 1 and the replica in datacentre 2 in a different location. If Datacentre 1 has a problem that knocks out the shared storage or all the nodes, then Datacentre 2 has a up-to-date copy to spin up and take over.

 

There isn't any advantage to have a replica server on the same storage as the primary because it provides no redundancy. 

 

All of the solutions (Checkpoint, failover cluster, replica) mostly make sense when you look at what scale of redundancy you require and how much you are willing to spend to achieve it. 

 

Hope this helps :)

 

Mark