http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=207907...
VMware said they did support it at first then they have since said no and reffered me to the fact that its highly experimental
I think you mix two things together. First, iirc vmfs5 has standard block-size 1MB (sub-block 8kB). AFAIK, you can not change this. Second thing is underlying hardware. Traditional disks can have 512B/4kB sectors, in case of SSD it is not important at all as sector-size is abstract (r/w ops are done in blocks).
So you can not create datastores WITH 4k-sectors, but you can create datastore ON disk with 4k-sectors...
As long as your disk/raid controller supports drives with 4KB sector-size you are OK on vSphere side. It should not be any problem for VMFS5 file-system because it uses block/sub-block size 1MB and ESXi will not be aware of RAID levels on the storage array. If you are presenting the LUNs from 4KB sector-size drives it should function normally.
Have you guys even tried 4k sector disks?
4k support is still far out on the roadmap...
http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=207907...
VMware said they did support it at first then they have since said no and reffered me to the fact that its highly experimental
I have one datastore on 4kB-sector drive for a year, it works without any problem. Maybe it is because most of 4kB-sector drives support 512B-sector emulation...
BTW, this all applies to SATA-drives only. SAS-drives can have different sector-size (520B, 528B, etc). Those extra-bytes are normally used rather for checksum not data, but this strange sector-size is still reported to OS...