VMware vSphere For Dummies by Dan Mitchell & Thomas Keegan

Author:Dan Mitchell & Thomas Keegan
Language: eng
Format: epub
Publisher: For Dummies
Published: 2011-02-06T16:00:00+00:00

Whenever you deal with storage, remember that storage is usually absolute. When you make changes at a low level to disks, there is no Recycle Bin that provides recovery. Be careful here, as most of the configuration in this chapter can profoundly modify data; if you perform an operation on an unintended target, you may not be able to recover from the damage.

vSphere storage options include

Virtual Machine File System (VMFS)

Network File System (NFS)

VMware Virtual Machine File System

In many operating systems, such as Microsoft Windows, you can’t natively share a file system across multiple hosts. Special configurations, such as clusters, create unique file system filters to allow multiple servers to see the same disk, but even then they usually don’t allow multiple servers to read and write to the same disk. vSphere uses Virtual Machine File System (VMFS), a proprietary clustered file system that allows many hosts to mount the same disk and use it. Any volume mapped to an ESX or ESXi host is called a datastore.

VMFS is used by many companies running vSphere. Custom-designed by VMware for virtualization, VMFS provides the best performance and features for storage and execution of VMs and snapshots of VMs.

VMFS allows up to 64 hosts to access the same volume. In practice, 64 hosts would be very unlikely because the hosts must contend with one another when performing some operations, such as power on and changes to a VM, so you usually don’t want so many hosts to have access to the same volume.

VMFS doesn’t have to actually lock the file system often. When it does, VMFS uses something called a Small Computer System Interface (SCSI) reservation. A SCSI reservation is a brief time when a single host has complete control of the volume — for example, vSphere issues SCSI reservations when you power up a VM or create, delete, or change the name or size of a file. Because thin-provisioned files grow, using thin provisioning can increase the amount of SCSI reservations. SCSI reservations can reduce performance, so it is wise not to put too many VMs on a single datastore. In vSphere 4.1, some arrays provide a plugin called a vStorage API for Array Integration (VAAI). A VAAI improves performance by offloading several functions to the array, and also reduces the need for SCSI reservations.

Because VMFS was designed for holding whole virtual machines, it supports large files and large volumes. Volumes can be up to 64TB , but they’re never that big because of performance and some limits by the SCSI protocol. The file size limit in VMFS is 262,144 blocks, and a block can be either 1, 2, 4, or 8MB. This translates to maximum file sizes of 256GB, 512GB, 1TB, and about 2TB. The 2TB size is very slightly less than 2TB, because 512 bytes is used by some metadata and the SCSI command set.

When you create a VMFS, you must choose the block size. Once you choose, you can’t change it without destroying and reformatting the volume, so choose wisely.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.