vSphere Storage Overview

Back to GigaNetworkers 

 

The vSphere Hypervisor, ESXi, provides host-level storage virtualization, which logically abstracts the physical storage layer from virtual machines.

A vSphere virtual machine uses a virtual disk to store its operating system, program files, and other data associated with its activities. A virtual disk is a large physical file, or a set of files, that can be copied, moved, archived, and backed up as easily as any other file. You can configure virtual machines with multiple virtual disks. To access virtual disks, a virtual machine uses virtual SCSI controllers. These virtual controllers include BusLogic Parallel, LSI Logic Parallel, LSI Logic SAS, and VMware Paravirtual. These controllers are the only types of SCSI controllers that a virtual machine can see and access.

Each virtual disk resides on a vSphere Virtual Machine File System (VMFS) datastore or an NFS-based datastore that are deployed on physical storage. From the standpoint of the virtual machine, each virtual disk appears as if it were a SCSI drive connected to a SCSI controller. Whether the actual physical storage device is being accessed through parallel SCSI, iSCSI, network, Fibre Channel, or FCoE adapters on the host is transparent to the guest operating system and to applications running on the virtual machine.
 
The vSphere storage management process starts with storage space that your storage administrator allocates on different storage systems prior to vSphere ESXi assignment. vSphere supports two types of storage - Local and Networked. Each type is detailed in the following lesson steps.

The illustration above depicts virtual machines using Local VMFS storage directly attached to a single ESXi host. Local storage can be internal hard disks located inside your ESXi host, or it can be external storage systems located outside and connected to the host directly through protocols such as SAS or SATA.
 
 

 
Networked storage consists of external storage systems that your ESXi host uses to store virtual machine files remotely. Typically, the host accesses these systems over a high-speed storage network. Networked storage devices are typically shared. Datastores on networked storage devices can be accessed by multiple hosts concurrently, and as a result, enable additional vSphere technologies such as High Availability host clustering, Distributed Resource Scheduling, vMotion and Virtual Machines configured with Fault Tolerance. ESXi supports several networked storage technologies - Fiber Channel, iSCSI, NFS, and Shared SAS.

 
 

Thick Provisioning

Thick provisioning is a type of storage pre-allocation. With thick provisioning, the complete amount of virtual disk storage capacity is pre-allocated on the physical storage when the virtual disk is created. A thick-provisioned virtual disk consumes all the space allocated to it in the datastore right from the start, so the space is unavailable for use by other virtual machines.
There are two sub-types of thick-provisioned virtual disks:

• A Lazy zeroed disk is a disk that takes all of its space at the time of its creation, but this space may contain some old data on the physical media. This old data is not erased or written over, so it needs to be "zeroed out" before new data can be written to the blocks. This type of disk can be created more quickly, but its performance will be lower for the first writes due to the increased IOPS (input/output operations per second) for new blocks;
• An Eager zeroed disk is a disk that gets all of the required space still at the time of its creation, and the space is wiped clean of any previous data on the physical media. Creating eager zeroed disks takes longer, because zeroes are written to the entire disk, but their performance is faster during the first writes. This sub-type of thick-provisioned virtual disk supports clustering features, such as fault tolerance.

For data security reasons, eager zeroing is more common than lazy zeroing with thick-provisioned virtual disks. Why? When you delete a VMDK, the data on the datastore is not totally erased; the blocks are simply marked as available, until the operating system overwrites them. If you create an eager zeroed virtual disk on this datastore, the disk area will be totally erased (i.e., zeroed), thus preventing anyone with bad intentions from being able to recover the previous data – even if they use specialized third-party software.

Thin Provisioning

Thin provisioning is another type of storage pre-allocation. A thin-provisioned virtual disk consumes only the space that it needs initially, and grows with time according to demand.

For example, if you create a new thin-provisioned 30GB virtual disk and copy 10 GB of files to it, the size of the resulting VMDK file will be 10 GB, whereas you would have a 30GB VMDK file if you had chosen to use a thick-provisioned disk.

Thin-provisioned virtual disks are quick to create and useful for saving storage space. The performance of a thin-provisioned disk is not higher than that of a lazy zeroed thick-provisioned disk, because for both of these disk types, zeroes have to be written before writing data to a new block. Note that when you delete your data from a thin-provisioned virtual disk, the disk size is not reduced automatically. This is because the operating system deletes only the indexes from the file table that refer to the file body in the file system; it marks the blocks that belonged to "deleted" files as free and accessible for new data to be written onto. This is why we see file removal as instant. If it were a full deletion, where zeroes were written over the blocks that the deleted files occupied, it would take about the same amount of time as copying the files in question. See the simplified illustration below.

 

Back to GigaNetworkers