This article demonstrates demonstrates how to configure disk redundancy VMware VSAN and is a follow on post from my How to enable VMware VSAN
The previous article showed how to configure a vSphere cluster for use with VMware VSAN.
Configure Disk Redundancy in VMware VSAN
When we configure Disk redundancy in VMware VSAN we do not setup any hardware based RAID configuration (remember the design consideration, this was a pre-requisite to have Pass-Through RAID Controllers). So how do we ensure data protection? VMware VSAN uses Storage policies to do this. These can be found in vCenter Home -> Rules and Profiles -> VM Storage Profiles.
Using VM Storage Profiles you can assign capabilities to storage. In a Cloud environment, we like to use Gold, Silver and Bronze to determine our storage capabilities. We can easily create these policies to now match the capabilities of the storage.
We have a number of different configuration options available to us when adding capabilities. The screenshot below shows these capabilities.
If we select Number of failures to tolerate and set this to 1, this specifies that the VMs associated with this policy can tolerate at least one failure (host, network or disk).
We have the option to set the Number of disk stripes per object, which specifies the number of hard disks the replica of a storage object is striped across.
The nice thing about this is immediately I can tell whether or not any datastores are capable of understanding the requirement in the matching resources window. The vsanDatastore shown below is capable of understanding these requirements that I have placed in the VM Storage Policy.
Note that this is no guarantee that the datastore can meet the requirements in the VM Storage Policy. It simply means that the requirements in the VM Storage Policy can be understood by the datastores which show up in the matching resouces.
Once Virtual Machines have been deployed, you can see how the placement has been performed. You can see this by looking at the VM Storage Policy under the manage tab on a VM.
The RAID 1 indicates that the VMDK has a replica. This is to tolerate a failure, the value was set to 1 in the policy. This allows the VM to continue to run if there is a single failure in the cluster. The witness is there to act as a tiebreaker. If one host fails, and one component is lost, then this witness allows a quorum of storage objects to still reside in the cluster. Notice that all three components are on different hosts for this exact reason.
By examining the lay out of the object above, we can see that a RAID1 configuration has been put in place by VSAN, placing each replica on different hosts. This means that in the event of a host, disk or network failure on one of the hosts, the virtual machine will still be available.
If the host on which the VM does not reside fails, then no action is required. If the host on which the VM resides fails, then vSphere HA can be used to automatically bring the VM online on one of the remaining hosts in the cluster.
More information on VMware VSAN can be found by clicking here