NFS or iSCSI for ESXi 5.5 and or 6 ??

[ Edited ]

We have NFS licenses with our FAS8020 systems. Will VMWare run ok on NFS, or should we revisit to add iSCSI licenses?  We are on Dell N4032F SFP+ 10GiB.  We have a different VM farm on iSCSI that is great (10GiB on Brocades and Dell EQs). Will NFS be as good or better performance and reliability wise? Thanks.

Re: NFS or iSCSI for ESXi 5.5 and or 6 ??

I prefer nfs, but I'm biased Smiley Happy The good news is we can run all protocols and you can test between the two.  This is a very old tech report but showed 9% delta between all protocols.  iSCSI had a higher cpu load on the esx host, but that may have changed since this paper was written.  The operational efficiencies of nfs are why I like it...dedup at the volume doesn't require thin provisioned luns to get the savings back, growing and shrinking is just simple by modifying the volume size, files can be dumped to tape or read without mounting vmfs.

Re: NFS or iSCSI for ESXi 5.5 and or 6 ??

Hello Brian,


You might want to give TR-4333 a read, it has a lot of best practices on using clustered Data ONTAP with vSphere.  


Performance is nearly equal, close enough to not really matter (it's even closer than the 9% in the TR mentioned with modern vSphere versions).  Reliablity is the's still a NetApp device providing the storage.  If you're already using iSCSI then your network team is already familiar with the change in operational proecdures needed to support storage traffic on an IP network, so there's no need to worry there (I have seen many times in historically FC organizations where the IP network team will reboot switches and do other things which cause network "blips" anytime...this is bad when storage uses the IP network).


With that being said, the choice between NFS and block is primarily about manageability in my opinion.  As was mentioned by @scottgelb, NFS datastores immediately show the benefits of storage efficiency (deduplication, compresson, thin provisioning) from both the NetApp and vSphere perspectives.  When using a block protocol, that efficiency is only visible to the NetApp administrator.  For example, if you use deduplication on a 1TB datastore that has 500GB of data which saves 50% of capacity, then the NFS datastore will show that 50% reduction immediately...the vSphere volume and NFS datastore would both show 250GB of capacity used.  With block, the VMFS datastore would still show the datastore being 50% used, but the NetApp volume would only have 250GB used.


Hope that helps.



If this post resolved your issue, please help others by selecting ACCEPT AS SOLUTION or adding a KUDO.