30 September 2013

Building Hyper-V Clusters Step by Step - Part IV - iSCSI Cluster Shared Volume (CSV) SCVMM

In part III of this series we managed to build a block storage Hyper-V failover cluster manually.  When you  are using SCVMM it will automate a lot of the steps for you.  For this to work you need to provision your storage through VMM.  Because we are using Server 2012 R2 it can perform everything directly from SCVMM 2012 R2.

Design Goals
  • Two node Hyper-V cluster
  • Using iSCSI as shared storage
  • Using two Cluster Shared Volume between 2 nodes
  • Using only single NIC per server
  • Using DHCP
  • Use SCVMM to configure everything
Servers
The same basic configuration as the previous labs

Cluster nodes HV01 and HV02
HV03 is the management server, iSCSI target and SCVMM Server

Build Process

Step 1 Add a storage provider

  • Fabric - Storage - Providers - Add Storage Devices
  • San and NAS devices descovered and managed by a SMI-S provider
  • Protocol: SMI-S WMI
  • FQDN of HV03 
  • Specify a Run As account
  • Discover should succeed and you should see and item in the list
  • Create a classification (2012 R2 iSCSI)
  • Select two drives to use and choose your classification
  • Finish the Wizard




Step 2 - Create logical Units
This is where you create iSCSI drives for the nodes to use.

  • Fabric - Storage - Classification and Pools
  • Select the Clasification created earlier
  • Click Create logical unit
  • Create one small logical unit (1GB - for the Quorum)
  • Create a bigger logical unit ( For the CSV)

At this stage it should look something like this



Step 3 - Create a host group
  • Fabric - Servers - All hosts - Create Host Group
  • View the properties of the host group
  • Select storage 
  • Allocate Storage Pools - Select and Add both
  • Allocate Logical Units - Select and add all three
  • Close properties
  • Add the Hyper-V hosts
  • Add the individual hosts as per normal
  • If you check the properties of the hosts look at storage
  • iSCSI Array should have the target specified 
Step 4 - Create the cluster
  • Fabric - Servers - All Hosts
  • Create - Hyper-V Cluster
  • Specify Name
  • Specify Run As account
  • Add the two nodes prepared earlier
At the storage screen you will notice that one of the disk is grayed out.  This will be used for the Quorum.  It selects the smallest partition for this.




  • For the other partition check all three boxes (Quick Format - Force Format - CSV)
  • Finish the wizard

Check progress on the Job as it can run for a good few minutes


Step 5  - Check Storage on the cluster
  • Fabric - Servers - All Hosts - Your Cluster Group
  • Open the properties of the cluster
  • Select Shared volumes
  • You should see the CSV here, if not then check available storage, select the volume and click the Convert to CSV button



Step 6 - Create a HA VM
  • VMs and Services - Create Virtual Machine
  • Create New VM
  • Specify a name
  • Configure hardware - Make sure you select availability and check "Make this virtual machine highly available"
  • Specify the host group as a destination
  • Select either host
  • The virtual machine path would be  C:\ClusterStorage\Volume1\   or similar - this refers to the CSV volumes
  • Finish the wizard
Conclusion
Using SCVMM you can drive the entire deployment of Hyper-V iSCSI fail-over clustering as long as your storage plays along nicely.  If you manually go and check the iSCSI initiators and the iSCSI targets on the individual servers you will notice that it is pretty much the same as when we completed lab III

No comments:

Post a Comment