New ESXi maximum limits, a completely new Virtual Machine File System and a space reclamation change highlight...
new storage features in vSphere 6.5. While the Virtual Machine File System is one of the bigger improvements, the change for space reclamation might be the most helpful.
Automatic Space Reclamation, known as UNMAP in previous versions of vSphere, is running automatically in the background; no manual trigger is needed. One feature that is gone is the Windows C# client, which was used to access the virtual infrastructure. Now the only way to access vCenter Server -- installed on Windows or as a Linux appliance -- is through the Flash-based vSphere Web Client. The new HTML5-based user interface doesn't implement all of the features.
VMFS 6 File System
VMware vSphere 6.5 supports up to 1,000 hosts connecting to the same data store with Virtual Machine File System (VMFS 6). In vSphere 6.5, ESXi hosts can support up to 2,000 paths in total, which is nearly double the limit from the previous version.
VMFS 6 improves snapshots, making things more efficient; snapshots have had issues in previous versions. In vSphere 6.5 Space Efficient Virtual Disks (SE Sparse Disk) are now the default virtual disks and are the snapshot format supported on VMFS 6. The VMFSparse snapshot format isn't supported in VMFS 6, but previous versions, such as VMFS 5, still do. Creating a quiesced snapshot in previous versions of vSphere often caused problems because of the time limit. Enhancements to vSphere Storage APIs allow admins to configure timeouts when creating a quiesced snapshot.
VMFS 6 introduces two new block sizes: small file block and large file block. The small file block size in vSphere 6.5 is set to 1 MB only. The small file block has a range from 64 KB to 1 MB; VMFS 6 uses a small file block of 1 MB only in this version. The large file block size is set at 512 MB in vSphere 6.5.
However, taking advantage of the new VMFS 6 may not be easy for every user. One main issue is that there is no in-place upgrade for VMFS 6. In order to upgrade -- which can be a tedious process -- follow these steps:
First, create a VMFS 6-based data store on new storage. Next, move the VMs away from VMFS 5 data store to the VMFS 6 data store via storage vMotion. Unmount the VMFS 5 data store in order to delete it and, finally, go to Add New Storage Wizard and create and format that with VMFS 6.
The fact that there is no automatic upgrade is a huge disappointment. It can be a time consuming process that not many users will want to handle.
Space reclamation with UNMAP
VMFS UNMAP is a command that was originally introduced in vSphere 5.5 that allowed users to reclaim unused storage blocks on a data store. VMFS UNMAP, now known as Automatic Space Reclamation, is used to reclaim what VMware calls "stranded space," also known as dead space. In vSphere 6.5, thin provisioned LUNs are automatically worked on. With the default settings on, Automatic Space Reclamation will take back any blocks that aren't being used and make them available to admins for storage. In previous versions, reclaiming unused space was only possible through the command-line interface; Automatic Space Reclamation will take back unused space in vSphere 6.5 without a manual prompt. Storage capacity is reclaimed and released to the array so other volumes can use these blocks when needed.
The process frees up deleted space from the back-end array in the background without any performance penalty. Automatic Space Reclamation also works at the guest OS level. If you delete files and folders within the guest OS, Automatic Space Reclamation shrinks the VMDK on a data store.
New, larger disk drives and 512e mode
VMware vSphere 6.5 introduces support for 4K native drives in 512 emulation (512e) mode. In order to provide drives with larger capacity, Advanced Format (AF) drives use a larger physical sector size of 4,096 bytes -- also known as 4K AF format. Disk drive vendors build high capacity drives with AF, which translates to better performance and larger drives that are cheaper. Storage devices that use 512e mode can also support legacy applications and guest OSes. Since most legacy applications can't support 4K native drives, 512e mode comes into play. The drives have a physical sector size of 4K but a logical sector size of 512 bytes in order to support legacy applications.
However, the AF drives in 512e mode are only supported on VMFS 6 formatted data stores. That is another reason for users to migrate to VMFS 6-formatted data stores.
How to configure vSphere I/O Control
What are the benefits of vSphere Storage I/O Control?
Your array may be key to fixing VMware storage issues