Backup of Cluster only 1 machine at a time?

[ravelin]

Hi
Below is an example of my problem:
I have a cluster with 2 hosts.
I have 4 machines, 2 on each host. (a,b,c,d)

I make a replication job for the cluster to a third standalone host.

However, only one machine is replicated at a time.
I want each host to process a maximum of 1 machine at a time, but always process if there is a guest that needs to be processed on that host.

Right now, when I start replication it starts with machine a, and the rest of the 3 are waiting for machine a to be done, before machine b starts and so on.
This way, one of my hosts will always be idle.

How can I minimise the time of replication creation and cpu usage per host?

[foggy]

However, only one machine is replicated at a time.

Kenneth, within a job, VMs are processed sequentially. If you want the VMs to be processed concurrently, put them in separate jobs and schedule the jobs accordingly. (I'm assuming here you have local storage on both hosts, do you?)

How can I minimise the time of replication creation and cpu usage per host?

Consider using the offhost backup mode to take data processing off the Hyper-V host. Please review the product user guide for details.

[ravelin]

Sorry, dont know why I was not notified of your response to the post.
Since its a cluster with a HP P2000 SAN, they share the same storage. I understand I can do each vm in a job by itself, its just not very flexible and I have to add new jobs each time someone adds a new guest on the cluster.
Will the offhost backup mode lower the time of creation of the replica happening every 15 minutes.

[foggy]

Since it shifts the replication load to a dedicated machine, eliminating unwanted overhead on the production host, this can also lower the time required to perform replication (given the offhost proxy possesses enough CPU resources). Also, since hardware provider is used to take a snapshot, the CSV stays in the redirected mode less time comparing to onhost processing.

However, make sure you deploy an offhost backup proxy on a host that is NOT a part of a Hyper-V cluster (see user guide for details).

[ravelin]

I tried to create the replica with standalone hosts.
That way I now have 2 jobs, "replicate hyperv1" and "replicate hyperv2". Both has the target host ReplicaHyperV3.
Now I start "replicate hyperv1" job, and moments later I start "replicate hyperv2".
"replicate hyperv1" Begins to replicate to ReplicaHyperV3.
"replicate hyperv2" says: Waiting for backup infrastructure resources availability

Cant I have more than 1 host replicating at the same time to my destination ReplicaHyperV3?

[Gostev]

What's max concurrent tasks setting in the ReplicaHyperV3 properties?

[chrisdearden]

If both hosts are using software VSS on the CSV , does that not only allow 1 backup per CSV at a time ?

[Gostev]

That could be another reason with software VSS and Hyper-V 2008. No such limitations if either Hyper-V 2012 or hardware VSS provider is used.

[ravelin]

It seems that it's because of software vss writer.
The first host who initiates a backup, locks the csv volume in redirected access.
Now I am troubleshooting why the IBM hardware vss provider is not working.
I have support in on it finally.