vSphere 6 finally bumped the version of NFS that was supported as a datastore from v3 to v4.1. NFS v4.1 is certainly not new having been introduced in 2010 but VMware has never seen to support v4.1 until vSphere 6. As expected NFS v4.1 brings a number of enhancements over v3 including support for multi-pathing and Kerberos authentication (AD) but there are a number of caveats and limitations with using it in vSphere 6 that you should be aware of:
- You cannot use Virtual Volumes (VVOLs) with NFS v4.1
- NFS 4.1 does not support hardware acceleration (VAAI) as a result you cannot create thick virtual disks on NFS v4.1 datastores (thin only) or use any of the VAAI-NAS primitives (i.e. Fast File Clone)
- According to the vSphere 6 Storage documentation on page 151, NFS 4.1 does not support the Fault Tolerance (FT) availability feature in vSphere, however on page 153 of that same documentation (see chart below) it is listed as supported. VMware needs to clarify this contradiction in their documentation.
[important]
Update from Cormac on this:
VMs on NFS v4.1 support FT, as long as it is the new FT mechanism introduced in 6.0.
VMs running on NFS v4.1 do not support the old, legacy FT mechanism.
In vSphere 6.0, the newer Fault Tolerance mechanism can accommodate symmetric multiprocessor (SMP) virtual machines with up to four vCPUs. Earlier versions of vSphere used a different technology for Fault Tolerance (now known as legacy FT), with different requirements and characteristics (including a limitation of single vCPUs for legacy FT VMs).
[/important]
- You cannot use v3 and v4.1 NFS versions to mount the same datastore as they do not use the same locking protocol and doing this cause data corruption
- NFS v3 and NFS v4.1 datastores can coexist on the same host
- Just like you can’t upgrade from VMFS3 to VMFS 5, you cannot upgrade an existing NFS v3 datastore to v4.1
- Be aware that vSphere 6 supports both NFS v3 and v4.1 but to do this ESXi has to use different NFS clients
- NFS v4.1 provides multipathing (pNFS) and you can use multiple IP addresses to access a single NFS volume
Below is a feature comparison chart that shows what vSphere features are not supported with NFS v4.1:
For more information on implementing NFS v4.1 read through the vSphere 6 Storage Documentation.
4 comments
Skip to comment form
Hi Eric,
Small error – misunderstanding: the multi-pathing in the NFS 4.1 that vSphere 6 brings is NOT pNFS but session-trunking. I explained this in an earlier blogpost here: http://hansdeleenheer.com/vsphere-6-nfs4-1-does-not-include-parallel-striping/
No judgment on VMware, just making sure we are all on the same path!
Hi Eric,
just point out another small correction, it is in fact possible to upgrade from VMFS3 to VMFS5 but you might be giving up some features depending on your current VMFS3 configuration (for example current block-size will remain the same after an upgrade)
Working on getting you a definitive NFS v4.1 & FT interop statement Eric – watch this space!
I think I have an answer for you Eric.
VMs on NFS v4.1 support FT, as long as it is the new FT mechanism introduced in 6.0.
VMs running on NFS v4.1 do not support the old, legacy FT mechanism.
In vSphere 6.0, the newer Fault Tolerance mechanism can accommodate symmetric multiprocessor (SMP) virtual machines with up to four vCPUs. Earlier versions of vSphere used a different technology for Fault Tolerance (now known as legacy FT), with different requirements and characteristics (including a limitation of single vCPUs for legacy FT VMs).