VMware vSphere Storage is the system that provides the central storage for all your virtual machines, separating their data from the physical server hardware. It matters because without storage, your VMs would have nowhere to keep their operating systems, applications, and user data, making the entire data center useless. For VCP-DCV, understanding how to configure and manage datastores and storage protocols is essential for building a reliable, scalable virtual infrastructure.
Jump to a section
A simple way to picture VMware vSphere Storage
When you move into a new apartment building, you first sign a lease with the building manager, which gives you permission to use a certain amount of space. This lease is like a storage protocol — it defines the rules for how you can access and store your belongings. Once the lease is signed, the manager assigns you a specific storage unit in the basement. That unit is your datastore — a formatted container where your virtual machines (VMs) will hold all their files, operating systems, and data.
Now, within that storage unit, you don't just throw everything in a heap. You organise your items using plastic boxes — each box is a virtual disk (VMDK) that holds a specific part of your VM, like its operating system or application data. The building manager might offer different types of storage space: some units are on a concrete floor (like a standard datastore on a local hard drive), while others are part of a shared, high-security vault accessible via a special key card (like a shared SAN storage using the Fibre Channel protocol).
When you need to add more shelf space, you don't move out — you just add another plastic box to your unit, or you ask the manager to enlarge your unit (like extending a datastore). If a fire breaks out in one part of the basement, your neighbour's storage unit can take over because the building manager automatically redirects access (like VMware's Storage vMotion or High Availability). The key point is: the lease (protocol) gives you the rules, the storage unit (datastore) gives you the space, and the plastic boxes (VMDKs) organise your data within that space.
Virtual machines are like computers inside a computer. They need storage, just like a physical computer needs a hard drive. But in a virtual environment, that storage doesn't have to be a single hard drive sitting inside the server. Instead, vSphere lets you pool storage from many different sources and present it to VMs in a flexible way.
At its core, vSphere Storage uses two key concepts: datastores and storage protocols. A datastore is a logical container that holds the files of one or more virtual machines. Think of it as a formatted filing cabinet. The datastore can reside on different types of physical storage — local hard drives inside the ESXi host, a Storage Area Network (SAN), or Network Attached Storage (NAS). The protocols are the languages that allow the ESXi host to talk to that storage.
The most common types of datastores in vSphere are:
VMFS (Virtual Machine File System): This is VMware's own high-performance clustered file system. It allows multiple ESXi hosts to read and write to the same datastore simultaneously, which is essential for features like vMotion (moving a running VM between hosts).
NFS (Network File System): A network-based protocol where the datastore lives on a NAS device. ESXi hosts access it over a standard Ethernet network.
vSAN: A software-defined storage solution that pools local disks from multiple ESXi hosts to create a single, shared datastore.
Storage protocols define how the ESXi host connects to the storage device. The main ones you need to know for VCP-DCV are:
Fibre Channel (FC): A dedicated, high-speed network using special cables and switches. It is very fast and reliable but also expensive.
iSCSI (Internet Small Computer System Interface): This uses regular Ethernet networks to carry SCSI commands. It is cheaper than Fibre Channel but can be slower if the network is not properly configured.
FCoE (Fibre Channel over Ethernet): It wraps Fibre Channel frames inside Ethernet packets, allowing you to use a converged network for both storage and regular traffic.
NFS: Already mentioned, it uses TCP/IP over Ethernet.
When you create a VM, you assign it virtual disks (VMDK files). These VMDK files are stored inside a datastore. Each VMDK appears to the VM as a SCSI or IDE hard drive. The VM doesn't know or care whether the underlying storage is on a local SSD, a SAN array, or a NAS box. That abstraction is the magic of vSphere Storage.
Why did VMware create this? In the old days, each physical server had its own hard drives. If a server failed, the data on those drives was hard to recover, and there was no easy way to move workloads. With shared storage (like a SAN), multiple servers can access the same data. If one server fails, another can take over instantly. This is the foundation of High Availability (HA), vMotion, and Distributed Resource Scheduler (DRS).
For beginners, the most important thing to grasp is the separation of storage from compute. The ESXi host is the compute part (the brains), and the datastore is the storage part (the memory and filing system). They can be separate physical devices connected by a network. This separation is what makes the advanced features of vSphere possible.
Let's take a concrete example. You have a company with five ESXi hosts. You buy a SAN array that contains 20 physical hard drives. You carve out a LUN (Logical Unit Number) from the SAN and present it to the five hosts via Fibre Channel. On that LUN, you create a VMFS datastore called "Production_Data". Now, you deploy four VMs onto that datastore. Because the datastore is shared, any of the five ESXi hosts can run any of the four VMs. If one host fails, vSphere automatically restarts those VMs on another host. That would be impossible if each VM's storage were locked to a single host's local disk.
In summary, vSphere Storage is about creating flexible, shared pools of storage that VMs can use, using file systems (datastores) and communication languages (protocols) to make it all work.
Plan Storage Architecture
Assess your requirements: how many VMs, performance needs, capacity, and budget. Choose whether to use block-level (FC, iSCSI) or file-level (NFS) storage. This step defines the protocol and datastore type.
Provision Storage on the Array
On your SAN or NAS device, create a RAID group from physical disks, then carve out a LUN (for block-level) or a shared folder (for NFS). Present the LUN to the ESXi hosts via zoning (FC) or target configuration (iSCSI).
Configure ESXi Host Storage Adapters
On each ESXi host, scan for new storage. For Fibre Channel, ensure the HBA is recognised. For iSCSI, enable the software iSCSI initiator, set the target IP, and perform a rescan. For NFS, specify the NAS share path.
Create a Datastore in vCenter
In the vSphere Client, navigate to the Storage view, choose 'New Datastore', select the LUN or NFS path, name the datastore, and choose VMFS (or NFS as the type). The wizard will format the storage and mount it on all hosts.
Deploy VMs onto the Datastore
When creating a new VM, select the newly created datastore as the location for the VMDK files. Alternatively, use Storage vMotion to migrate existing VMs to the new datastore without downtime.
Monitor and Maintain the Datastore
Use vCenter to monitor capacity, performance, and health. Set alarms for when the datastore reaches a certain percentage of fullness. Periodically perform storage maintenance like unmap operations to reclaim space.
Imagine you are an IT administrator for a medium-sized company with 50 employees, all using virtual desktops and applications hosted on a vSphere cluster. Your boss comes in and says the company is moving to a new building and needs to upgrade the storage because the old NAS is out of warranty. This is a perfect real-world scenario to apply vSphere Storage concepts.
First, you would assess the current storage. You log into vCenter Server and check the existing datastores. You see one datastore called "Legacy_NAS" using NFS protocol, which is 90% full. You also see several VMFS datastores on a small SAN that is almost out of space. Your first step is to plan the new storage architecture. You decide to buy a new all-flash SAN array that provides both Fibre Channel and iSCSI connectivity.
The next step is to physically install the SAN in the new data center (or the same building, but let's say you are moving). You connect it to your Fibre Channel switches (if using FC) or to your Ethernet switches (if using iSCSI or NFS). You then configure the SAN: you create a RAID group from the physical disks, and from that RAID group you carve out a LUN of 2 TB. You present this LUN to all seven ESXi hosts in your cluster.
Now, you need to configure the ESXi hosts to see the new storage. For Fibre Channel, this involves scanning the HBA (Host Bus Adapter) on each host. For iSCSI, you have to configure the software iSCSI adapter with the target IP address. For NFS, you specify the NAS server and share path.
Once all hosts can see the new LUN, you create a datastore on it. In vCenter, you right-click on the datastore view and choose "New Datastore". You select VMFS as the file system, give it a name like "Flash_SAN_01", and choose the LUN. The wizard formats the LUN with VMFS and mounts it on all hosts.
Finally, you need to move your VMs from the old datastores to the new ones. You use Storage vMotion, which allows you to migrate a running VM's files without downtime. You select the old VMDK files and choose the new datastore as the destination. The migration happens live, with no interruption to users.
The benefit? Your new all-flash array is much faster, making applications snappier. You also have redundancy because the new SAN has dual controllers. You have practised proper storage life-cycle management: decommissioning old hardware, provisioning new storage, and migrating workloads. In a real job, you would also document everything, create alert thresholds for capacity, and schedule regular performance monitoring.
The VCP-DCV exam (specifically exam 2V0-21.20 and similar) loves to test storage concepts in practical scenarios. You will see multiple-choice and multiple-select questions that require you to know the differences between storage protocols, the characteristics of VMFS vs. NFS, and the steps for provisioning storage. Here is what you must focus on.
First, memorise the key features of VMFS. It is a clustered file system that allows concurrent read/write access from multiple ESXi hosts. It supports features like vMotion and HA because of that shared access. A common trap question: "Which file system supports multiple ESXi hosts accessing the same datastore simultaneously?" The correct answer is VMFS. Beginners often confuse NFS here because NFS also supports multiple hosts, but NFS is a protocol, not a file system created by VMware. The exam will ask specifically about VMware's own file system.
Second, understand the difference between block-level and file-level storage. Block-level (Fibre Channel, iSCSI, FCoE) presents raw storage devices (LUNs) that ESXi formats with VMFS. File-level (NFS) presents a fully formatted file system (like a shared folder) that ESXi uses directly. A typical exam question: "Which of the following is a file-level storage protocol?" Answer: NFS. "Which is a block-level storage protocol?" Fibre Channel, iSCSI, FCoE.
Third, be clear on LUN sizes and VMFS properties. VMFS can span multiple extents if a LUN is too small, but the maximum single extent size is 64 TB. Also, VMFS datastores can be expanded on the fly by adding a new extent or by resizing the LUN if the array supports it.
Fourth, know the hardware compatibility. The exam will test that Fibre Channel requires specific HBA drivers and that iSCSI can use software (CPU-based) or hardware (dedicated adapter) initiators. A trap: asking which iSCSI initiator is used for the lowest CPU overhead — the answer is the hardware iSCSI adapter.
Fifth, Storage vMotion is a favourite topic. They will ask what is required: shared storage, a VMFS or NFS datastore, and that the VM is powered on. There is a trick question where they say "Storage vMotion requires the VM to be powered off" — that is false. Storage vMotion works with powered-on VMs with zero downtime.
Sixth, understand the difference between Raw Device Mapping (RDM) and VMDK. RDM gives a VM direct access to a LUN, bypassing the VMFS file system. It is used for legacy applications that need to see the actual disk hardware or for SAN snapshots. The exam tests that RDM can be in physical or virtual compatibility mode.
Finally, remember that NFS versions matter: vSphere supports NFS 3 and NFS 4.1. NFS 4.1 adds support for Kerberos authentication and multipathing. The exam may ask which version is required for secure authentication.
Common exam traps:
Thinking that all protocols require dedicated hardware. iSCSI can run over standard Ethernet.
Assuming that VMFS is the only file system for datastores. vSAN has its own distributed file system.
Confusing the roles of the initiator (ESXi host) and target (storage array).
Memorise these definitions precisely:
Datastore: A logical container that holds VM files, formatted with a file system (VMFS or NFS).
LUN: A logical unit of storage presented from a SAN array to a host.
VMDK: A virtual disk file that represents a hard drive to a VM.
HBA: A physical card (or software driver) that enables a host to connect to a storage network.
A datastore is a logical container that holds VM files and is formatted with either VMFS (for block-level storage) or NFS (for file-level storage).
Storage protocols define the communication method: Fibre Channel, iSCSI, and FCoE are block-level; NFS is file-level.
VMFS is a clustered file system allowing multiple ESXi hosts to read and write to the same datastore simultaneously, enabling vMotion and HA.
Storage vMotion migrates a VM's files between datastores while the VM remains powered on, with zero downtime.
Raw Device Mapping (RDM) allows a VM direct access to a LUN, used for specific compatibility, not general performance.
iSCSI can run over a standard Ethernet network, making it a cost-effective alternative to Fibre Channel.
NFS version 4.1 supports Kerberos authentication and multipathing, unlike NFS version 3.
A LUN is a logical unit of storage presented from a SAN array to a host, which is then formatted with a file system.
These come up on the exam all the time. Here's how to tell them apart.
VMFS
Clustered file system developed by VMware for block-level storage.
Allows concurrent read/write from multiple ESXi hosts.
Requires a LUN to be presented from a SAN array.
NFS
A file-level network protocol for accessing files on a NAS.
Allows read/write from multiple hosts over TCP/IP.
Does not use LUNs; mounts a shared folder from a NAS.
Fibre Channel (FC)
Requires dedicated Fibre Channel switches and HBAs.
Typically higher performance and lower latency.
More expensive due to specialised hardware.
iSCSI
Runs over standard Ethernet networks using TCP/IP.
Can be more cost-effective, but performance depends on network quality.
Can use software initiators (CPU) or hardware adapters.
Block-Level Storage (FC, iSCSI, FCoE)
Presents raw storage blocks (LUNs) to the host.
Host must format the LUN with a file system (e.g., VMFS).
Offers fine-grained control over I/O operations.
File-Level Storage (NFS)
Presents a fully formatted file system (e.g., NFS share).
Host accesses pre-existing files and folders.
Simpler management, as the NAS handles the file system.
Standard VMDK on VMFS
Stored as a file on a VMFS datastore.
Supports features like snapshots, vMotion, and thin provisioning.
Best for general-use virtual machines.
Raw Device Mapping (RDM)
Maps a LUN directly to a VM, bypassing VMFS.
Used for SAN-level snapshots or legacy applications that need direct disk access.
May not support all vSphere features (e.g., snapshots in physical compatibility mode).
Mistake
VMFS and NFS are both file systems.
Correct
VMFS is a file system (developed by VMware), while NFS is a protocol for accessing files over a network. When using NFS, the underlying file system is on the NAS device, not managed by VMware.
Both terms relate to storing data, so beginners often treat them as equivalent, but they serve different layers.
Mistake
All ESXi hosts must use the same storage protocol.
Correct
A single ESXi host can have multiple storage adapters and can connect using different protocols simultaneously (e.g., Fibre Channel for some datastores and NFS for others).
People assume uniformity because the host knows which storage it uses, but vSphere is designed for heterogeneous environments.
Mistake
A datastore can only hold files from one virtual machine.
Correct
A single datastore can hold the files (VMDKs, configuration files, snapshots) from many virtual machines.
The word 'store' might imply one customer, but in IT, a datastore is a shared multitenant container.
Mistake
Fibre Channel storage is always faster than iSCSI.
Correct
Performance depends on many factors including network infrastructure, disk type (SSD vs. HDD), and workload. In a correctly configured environment, iSCSI over 10GbE can match or exceed older Fibre Channel speeds.
Legacy wisdom and marketing create a hierarchy that isn't always true in modern data centers.
Mistake
Storage vMotion migrates the entire virtual machine, including its compute state.
Correct
Storage vMotion only migrates the virtual machine's files (VMDKs, configuration) from one datastore to another while the VM runs. vMotion migrates the running memory and CPU state to another host.
The similar names cause confusion between two distinct operations.
Mistake
Raw Device Mapping (RDM) gives the VM better performance than a VMDK on VMFS.
Correct
In most cases, a VMDK on VMFS provides comparable or better performance. RDM is used for specific compatibility reasons (e.g., SAN snapshots, clustering), not for general performance.
The idea of 'direct access' sounds faster, but VMFS has optimisations that often make it faster.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
A datastore is a logical container that holds VM files, formatted with a file system (like VMFS). A storage protocol is the method (like Fibre Channel or iSCSI) used to connect the ESXi host to the storage device.
Yes, you can. A local datastore uses disks physically inside the server. However, it does not allow vMotion or HA between hosts because only that one host can access the drives.
A LUN is a logical unit of storage presented from a SAN array to a server. A datastore is a formatted file system created on that LUN (or on a NAS share). You create a datastore on top of a LUN.
It is strongly recommended, but not strictly required. Using the same network as regular traffic can cause congestion and poor performance. Best practice is to use dedicated VLANs or physical switches for iSCSI.
VMFS is VMware's clustered file system. It allows multiple ESXi hosts to read and write to the same datastore at the same time. This is essential for features like vMotion, HA, and DRS.
Yes, you can expand a VMFS datastore by adding a new extent (if the SAN allows) or by increasing the LUN size on the array and then using the 'Expand Datastore' wizard in vCenter. This is an online operation.
A VMDK (Virtual Machine Disk) file is a file that acts as a hard drive for a virtual machine. It is stored inside a datastore and appears to the VM as a physical SCSI or IDE drive.
You've finished VMware vSphere Storage. Continue through the VCP-DCV study guide to build a complete picture of the exam.
Done with this chapter?