VMware’s vSphere and Microsoft's Hyper-V are the two heading stages for server virtualization. Despite the fact that they give comparative gimmicks and usefulness, the way they oversee capacity for virtual machines is altogether different. VMware’s VSphere is a type 1 hypervisor and is based on a microkernel used to run features that are need to support the virtualization. Type 1 hypervisors run straightforwardly on server fittings and go about as the deliberation layer between the physical assets of the server and the virtual assets relegated to the virtual machine. Hyper-V is conveyed in two structures, either as a standalone Type 1 hypervisor (known as Hyper-V Server 2012 R2 in the most recent discharge) or as a component of the Windows Server working framework, where the Hyper-V gimmick is actualized as a "part". On both stages, the hypervisor oversees physical capacity assets and the presentation of capacity to the virtual machines. A virtual machine is embodied in various documents that speak to both the arrangement and substance of the virtual server. Hyper-V and the vsphere stage utilize the idea of a virtual hard plate, which is similar to the physical hard drive in a standard server. VSphere utilizes the VMDK (virtual machine circles) design, though Hyper-V uses VHD and VHDX (virtual hard circle and virtual hard plate amplified). Both stages additionally utilize various extra documents to track virtual machine design and to handle things when virtual machines are suspended and not dynamic and for depictions.
In the most recent discharges, vSphere (Esxi 5.5) backings VMDK records up to 62tb in size and Hyper-V backings VHDX documents up to 64tb. Both configurations permit the capacity for a virtual plate to be completely or slender provisioned at portion time. Vmdks "develop" by including VMDK document degrees. Hyper-V element Vhds develop by expanding the span of the VHD document itself. Microsoft proposes utilizing the VHDX design on the grounds that it is more space-effective and has extra peculiarities, for example, inserted metadata to relieve debasement issues. In both hypervisor situations, virtual machines need to be put away on a record framework. This is attained in marginally diverse routes on every stage. VSphere stores virtual machine records in datastores, which can be upheld on either NFS or square based stockpiling. Piece stockpiling can be exhibited through iscsi, Fiber Channel or Fiber Channel over Ethernet conventions and is designed with VMFS or VMware Virtual Machine File System. VMFS is a VMware restrictive record framework that gives gimmicks, for example, bunching, locking and depictions. It has developed over various cycles that take after the arrivals of vSphere. Hyper-V can use stockpiling through SMB (CIFS) shares utilizing Hyper-V for SMB, which was presented in Windows Server 2012. On the other hand, Hyper-V can utilize square stockpiling displayed to the server utilizing iscsi, Fiber Channel and Fiber Channel over Ethernet. Piece stockpiling can be organized utilizing either Windows NTFS or Refs record frameworks.
Vmdks "develop" by including VMDK document degrees. Hyper-V element Vhds develop by expanding the extent of the VHD record itself. Microsoft prescribes utilizing the VHDX position in light of the fact that it is more space-effective and has extra peculiarities, for example, inserted metadata to relieve debasement issues. Refs (Resilient File System) was presented by Microsoft with Windows Server 2012 and is a stronger and adaptable document framework focused around NTFS. One of the profits of utilizing NTFS or Refs is the capability to skim virtual machine documents specifically on the volume. vsphere ESX 5.5 hosts are constrained to a most extreme of 2,048 virtual plates. Additionally, outer stockpiling is restricted to 256 Fiber Channel Luns (with a most extreme LUN size of 64tb), 256 iscsi Luns or 256 NFS mounts.
Hyper-V hosts are obliged by the cutoff points of the Windows working framework, which scales to a hypothetical breaking point of numerous a great many Luns for every HBA thus, in down to earth terms, more than will ever be required for a solitary framework. Windows Server 2012 presented information deduplication for NTFS record frameworks. This was thusly improved in Windows Server 2012 R2. Microsoft helps the organization of virtual hard plates for VDI stockpiling workloads on Windows 2012 utilizing deduplication, which can bring about critical limit reserve funds. At this point, nonetheless, general server virtual machines are not underpinned, despite the fact that there are no specialized limitations to keep their usage.
Both vSphere Esxi and Hyper-V have the capacity use neighborhood stockpiling assets. In vSphere, neighborhood plates are organized with the VMFS record framework. VMware as of late presented the Virtual SAN characteristic, which permits three or more Esxi servers to go about as an imparted stockpiling group. Virtual machines are recreated on nearby stockpiling between the groups for flexibility and oblige the utilization of blaze stockpiling in every server to go about as a read/compose store. In outline, the hole away backing between Hyper-V and vSphere has contracted altogether. Very nearly all peculiarities are upheld similarly on both stages, while Hyper-V edges ahead on specific gimmicks, for example, information deduplication and adaptability for outside capacity. VMware is attempting to change the amusement by presenting Virtual SAN and moving some stockpiling go into the server, however Microsoft might effectively have the capacity to copy this gimmick through Storage Spaces, which would level the playing field once more.

Sources
http://www.computerweekly.com/feature/VMware-vs-Hyper-V-storage-comparison