locked
SCVMM and Hyper-V upgrade RRS feed

  • Question

  • Hi,


    We are starting an upgrade project for our SCVMM and Hyper-V infrastructure and I'm posting this in the hope someone has experienced a similar situation and has some advice. Our current environment is:

    Six Windows Server 2012 R2 Hyper-V Clusters, three nodes in each cluster.

    Two sites, each site has 3 clusters with Hyper-V replication used to replica VMs from main site to other site. This other site would be used in a DR fail-over situation.

    A single physical SCVMM 2012 R2 Server, managing all clusters. This server is based in the main site.

    The main production cluster is being replaced with new hardware and the original plan was to go with Windows Server 2019 Datacenter, also replacing the SCVMM server with a new physical SCVMM 2019 Server. This plan has changed now as SCVMM 2019 will not support the remaining clusters as they are 2012 R2 (SCVMM 2019 will only support 2016 or 2019 nodes). We are OK with this but is it a good idea to stick with a single SCVMM Server?.

    I had read the Microsoft SCVMM High Availability documentation but this seems overkill for our environment. If there is an outage with the SCVMM node then I'm sure this will be an acceptable risk as it will not affect the running nodes and VMs. It was mentioned that we could run the SCVMM Server as a VM on the new cluster but I'm more comfortable with having this as a separate physical server. This Server will be backed up so worst case scenario for a failure of SCVMM is a restore of SCVMM but a DR fail-over to the 2nd site would leave use without a SCVMM server.

    Any feedback on the above is welcome.

    thanks

    Sunday, August 2, 2020 6:40 PM

All replies

  • Hello,

    If you have the hardware which allows to install the SCVMM server cluster, you can implement it. If not, it's still acceptable because the failure of SCVMM server doesn't affect the running of Hyper-V server clusters and VMs.

    From my own opinion, it depends on the budget of the hardware investing.

    Besides, if you deploy the SCVMM server on the standalone physical machine, you still can deploy a SCVMM server in a VM on the Hyper-V clusters in standby. If there is a outage for the physical machine, you still can migrate the workloads to the VM.

    Best regards,

    AndyLiu



    "Virtual Machine Manager – General" forum will be migrating to a new home on Microsoft Q&A!
    We invite you to post new questions in the "Virtual Machine Manager – General" forum's new home on Microsoft Q&A!
    For more information, please refer to the sticky post.
    Monday, August 3, 2020 1:45 AM
  • Hi Andy,

    thanks. Is there any documentation that covers the scenario of having a VM in standby mode?

    Monday, August 3, 2020 3:31 PM
  • Hello,

    I'm sorry. There is no such documentation.

    Best regards,

    Andy Liu


    "Virtual Machine Manager – General" forum will be migrating to a new home on Microsoft Q&A!
    We invite you to post new questions in the "Virtual Machine Manager – General" forum's new home on Microsoft Q&A!
    For more information, please refer to the sticky post.

    Friday, August 7, 2020 8:47 AM
  • This is a question right up the same alley.  Hello, After we had a failed in-place scvmm 2016 to 2019 upgrade where vmm server would not uninstall we ended up with a very unstable 2016 scvmm. I have a new separate 2019 scvmm setup and patched and fully working, but empty. The question I have after researching the "reassociate this host with this VMM environment" check box is; 1. The new scvmm is 2019 and the servers still have the 2016 VMM agent installed. Does the reassociate check box upgrade the agent also and is there a reboot required for all nodes in the cluster? 2. The hosts are already clustered, full production and populated. Has there been any known issues bringing the whole cluster into a new VMM with the reassociate check box? 3. To remove the cluster from the 2016 SCVMM is the selection to close the "Remove" selection when right clicking on the cluster, removing a node removes the node from the cluster and I'm sure we don't want to uncluster. Current configuration is 2016VMM with 2-multi-node clustered 2016 servers with lots of VMs. Thanks in advance.

    Dave Kreitel


    Tuesday, August 18, 2020 3:53 PM