How to open Labview 2012 in 2009

Wolfgang on the Road

294: Mini "Proof of Concept" with Windows Server 2016 Storage Replication

September 13, 20151 Comment

One of the most interesting features of Windows Server 2016 is the synchronous or asynchronous replication of hard disk contents.

The possible scenarios and also a guide for installation are described here:

Actually, all scenarios are interesting. I find the replication in a "stretched cluster" particularly chic for use as a Hyper-V platform or as a normal file server.

The structure described in the Technet is much too complex "for playing". In this post everything is reduced to a minimum. The scenario “Hyper-V stretched cluster” is set up. The idea here is to expand a virtualization infrastructure over two locations and to replicate the storage via storage replication (SR).

Technical questions / conditions had to be tried out in practice beforehand due to the lack of clear documentation. Here are the most important rules of the game for the POC environment (built from Hyper-V VMs)

  • Hard disks built directly into the cluster nodes cannot be used for SR. That is why iSCSI targets are still used in this environment.
  • "Storage Spaces Direct", also a new Windows Server 2016 feature, are also ruled out because at least 4 machines would be required on each side to map the storage.

This resulted in this configuration based on Technology Preview 3:

  • TDC01 is DC and server for two iSCSI targets.
    Two virtual disks are configured in each target
  • TC01 and TC02 form a failover cluster (no special configuration,
    each node has only one network card).
  • All machines are in the same AD site.

The aim is to replicate the 100GB volume with the drive letter E: from TC02 to TC01 synchronously. The starting point for the following pictures is a configured failover cluster. The local disks are not yet assigned to the cluster.

The first step is to configure the necessary hard drives. Both sides are set up identically. The E: and L: drives are each partitioned as GPT and a volume is set up. E: simulates the data drive and L: the log disk for replication.

Now the disks are added to the cluster.

It is important to note down the numbers of the cluster disks in order to avoid confusion later. In this case, “Cluster Disk 4” will be the source and “Cluster Disk 3” the target of the replication (this can be seen from the host name in the Disk Info field).

The source disk (in this case "Disk 4") must now be converted into a cluster shared volume or added. The disc must be online for this. This only works if you do it on the cluster node to which the disk is connected. The action shown in the following picture will therefore fail (because the owner node is the wrong one).

This can be corrected simply by moving the disk to the other node (where the disk is also located).

Then you can take "Disk 4" online and generate the CSV (there is no separate picture for this). Only the source data disc is important for the "online" status. Everything else is irrelevant and will be correctly recognized and configured later by the wizard.

Now it's time to set up replication. The following pictures show the wizard step by step and are actually self-explanatory. Start with the right mouse button on "Cluster Disk 4".

Now the initial synchronization starts, which you can see very nicely in the Task Manager and Eventlog.

In the Cluster Administrator it now looks like this:

I had a few unsuccessful attempts until it was reasonably understandable. The most important here:

  • it happened that no disks could be added to the cluster. The reason is that there were still SCSI reservations from unsuccessful attempts. Help "Clear-ClusterDiskReservation -disk 1" and "Clear-ClusterDiskReservation -disk 2" is best on both sides.
  • Some of the cluster disks do not come online -> Windows Search Service must be deactivated.
  • When setting up the SR no target disk is displayed and a click on “non eligable disks” shows “Partition size does not match”. The reason for me was that I thought that the target disk should not or should not be partitioned. But must 🙂
  • And sometimes a reboot also helped.

You can also read all of this here: Storage Replica: Know Issues (Worth a look ;-))