vSphere backup and recovery is the safety net for an entire virtualised data centre. Without it, a single hardware failure, accidental deletion, or ransomware attack can erase months of work and bring a business to a standstill. For the VCP-DCV exam, you must understand not just that backups happen, but how to design a strategy that guarantees data can be restored when everything goes wrong.
Jump to a section
A simple way to picture vSphere Backup and Recovery
Have you ever spent a whole Sunday cooking a huge batch of your signature chilli, only to realise you cannot possibly eat it all before it goes bad?
You portion it into freezer bags, label each one with the date and contents, and stack them in your chest freezer. That freezer is your primary storage – it holds your live data, readily accessible. But what if the freezer breaks? What if the power goes out for a week? Everything melts, and you lose all that work.
That is why you also have a backup plan. Maybe you keep a separate, smaller freezer in the garage (that is your backup location – a different storage system, ideally in a different place). Or you might take one precious bag of chilli to your friend’s house down the street (that is an off-site backup – protecting against a disaster like a house fire that destroys both freezers).
But simply having the food in another freezer is not enough. You need to check, every so often, that the garage freezer is still cold, that the bag hasn't sprung a leak, and that the chilli is still edible. In IT, that is called a backup validation test. The vSphere Backup and Recovery tools do this automatically: they give you a guaranteed way to restore your ‘chilli’ – your virtual machines – quickly and fully, whether you need a single file (a single spoonful) or the entire server (the whole batch).
At its simplest, a backup is a copy of your virtual machine (VM) – all its files, operating system, applications, and data – stored someplace safe. But vSphere is not a simple desktop computer. A single VMware ESXi host might run dozens of VMs. A vCenter Server manages hundreds of hosts. So how do you back up an entire virtual infrastructure without causing chaos?
The answer is that vSphere has three main approaches to backup, and the VCP-DCV exam expects you to know when to use each one.
Image-level backup. This is the most common method for production environments. An image-level backup takes a snapshot of the entire VM disk (the VMDK – Virtual Machine Disk file) while the VM is still running. The backup software – called a backup agent or backup appliance – talks to vSphere through an API (Application Programming Interface). The API tells vCenter to quiesce the VM, which means it pauses the file system and flushes all pending write operations to disk so that the backup is crash-consistent (exactly as if the power were pulled). Then the backup software reads the VMDK file and copies it to a backup target – this could be a dedicated backup server, a storage area network (SAN), a network-attached storage (NAS) device, or cloud storage like Amazon S3. The advantage of image-level backup is that it captures the entire VM in one go. To restore, you simply deploy that image as a new VM. Disadvantages are that it uses a lot of storage space and takes time to back up large VMs.
File-level backup. Sometimes you do not need to restore a whole server – you just need one accidentally deleted spreadsheet. File-level backup works inside the guest operating system (the OS running inside the VM, like Windows Server or Linux). The backup software uses an agent installed inside the VM to pick individual files and folders, just like copying photos from your phone to an external drive. The benefit is granularity: you can restore a single document without touching the rest of the VM. The downside is that you must manage agents on every VM, and it does not protect against a full OS corruption.
vSphere Replication. This is not a traditional backup – it is a disaster recovery (DR) tool. vSphere Replication creates a live, ongoing copy of a VM’s data to a secondary site, typically at a different physical location. The replica VM is kept in a powered-off state. If the primary site goes down (say, a flood takes out the data centre), you power on the replica VM at the secondary site and business continues with only a few minutes of data loss. The VCP-DCV exam distinguishes backup (point-in-time copies you keep for months) from replication (near-real-time copies you use for quick recovery).
Key technical terms you must know:
RTO (Recovery Time Objective): The maximum amount of time you can afford to be without the service. If a VM must be back within 4 hours, your backup strategy must support that speed.
RPO (Recovery Point Objective): The maximum amount of data you can afford to lose measured in time. If your RPO is 1 hour, you must take backups at least every hour.
Backup window: The time period during which backups can run without affecting production performance. If your application is busy 9am-5pm, your backup window is overnight.
Deduplication: A technique that eliminates duplicate blocks of data across multiple VMs to save storage space. For example, if ten VMs all run the same version of Windows, deduplication stores that data only once.
Compression: Shrinking the backup file size to use less storage and network bandwidth.
vSphere offers a feature called VMware vSphere Storage APIs – Data Protection (VADP) – this is the framework that backup software uses to interact with vSphere. VADP enables efficient, agentless backups (no need to install software inside every VM) and supports both full and incremental backups (backing up only the changes since the last backup).
Without a backup strategy, a corrupted database or a ransomware attack would mean total data loss. The exam tests your ability to match the right backup method to the business requirement. For example, if the question says ‘a financial application requires recovery of individual transactions from 30 days ago,’ you need file-level backup. If the question says ‘the whole VM must be restored in under 15 minutes,’ you need image-level backup with fast storage. If the question says ‘we must have zero-minute failover to another data centre,’ you need vSphere Replication with an RPO of seconds.
In summary, vSphere backup and recovery transforms raw data into a recoverable asset. It replaces the old world of tape libraries and manual, error-prone scripts with automated, policy-driven, and centrally managed operations. For the VCP-DCV exam, you must be able to read a business requirement and pick the correct backup type, understand RTO/RPO, and know the role of VADP and vSphere Replication.
Define the Business Requirements
Meet with stakeholders to determine RTO (how fast you need to restore) and RPO (how much data loss is acceptable). Write these down – they drive every subsequent decision.
Select the Backup Method
Choose between image-level backup (for whole VM restore), file-level backup (for granular file recovery), and vSphere Replication (for disaster recovery to another site). Base the choice on the RTO/RPO and the VM’s role.
Choose Backup Storage
Decide where backups will be stored – a dedicated NAS/SAN on site for speed, plus cloud storage for off-site protection. Ensure the storage is in a different location from the production VMs to avoid a single point of failure.
Configure the Backup Job
In your backup software (e.g., Veeam), create a backup job that includes the target VMs, schedule (matching the backup window), retention policy (how many days to keep backups), and enable deduplication and compression to save space.
Perform a Test Restore
After the first backup completes, perform a restore to an isolated network. Verify that the VM boots, the data is consistent, and the recovery time meets the RTO. Document the test and repeat the process quarterly.
You are the new IT administrator for a mid-sized company called SwiftRetail, which runs its entire e-commerce platform on vSphere. The platform consists of 20 VMs: a web server, an application server, a SQL database, and several helper services. The CEO has just told you: ‘We cannot afford to lose more than one hour of orders, and if the system goes down, we need it back within 4 hours.’
Your first step is to formalise these numbers into RPO (1 hour) and RTO (4 hours). You then design a backup strategy that meets these requirements.
You install a VADP-compatible backup appliance (such as Veeam Backup & Replication – common in real environments) and connect it to your vCenter Server. You create a backup job for all 20 VMs. For the SQL database VM, you configure it for file-level backup every 30 minutes because the transaction logs are critical. For the web server and other less critical VMs, you schedule image-level backups every 2 hours. All backups are stored on a dedicated NAS device in the same data centre. You also set up a secondary backup copy to a cloud storage provider (like Amazon S3) for off-site protection.
You schedule the backup window for 2am-6am because website traffic is lowest. During the first backup, you notice the network is saturated and the web server is slow. You solve this by enabling network throttling in the backup software and using deduplication to reduce the data volume.
After a month, you run a test restore. You pick the SQL VM, restore it to an isolated network, and verify the database contents are intact and within your RPO. This test is documented and presented at the quarterly IT review.
One Tuesday, a developer accidentally runs a destructive script against the database. You go to the backup appliance, browse the restore points from 45 minutes ago, and restore the SQL database as a new VM. You attach the database to the existing application server, test the connection, and the e-commerce platform is back live in 20 minutes – well within the 4-hour RTO.
Later, the data centre suffers a cooling failure. You activate your disaster recovery plan: you have vSphere Replication replicating the critical SQL VM to a secondary site 50 miles away. You power on the replica there and update DNS. The order processing is back online in 15 minutes.
In this real scenario, the IT professional’s role is not just to ‘do backups’ but to:
Define RTO and RPO with business stakeholders.
Select the appropriate backup method (image-level, file-level, replication) per VM.
Configure backup schedules, retention policies (how long to keep backups – typically 30 days), and off-site copies.
Monitor backup job success/failure logs.
Test restores regularly (at least quarterly).
Document the entire process so that anyone on the team can perform a restore.
A common trap for beginners is to set up backups and then forget about them. The exam will test that you understand backup validation is a continuous process, not a one-time setup.
The VCP-DCV exam (specifically for objective 5.2) focuses on your ability to implement backup and recovery strategies. You will see scenario-based multiple-choice questions. The exam does not ask you to memorise specific backup software commands – it tests concepts and design decisions.
Here are the exact concepts you must know cold:
RTO vs RPO definitions: You will almost certainly get at least one question asking you to identify which is which. Trap: the question gives a scenario describing time and data loss, and asks ‘what is the RPO?’ The answer is always the amount of data loss in time, not the recovery speed.
VADP purpose: Questions will ask what VMware technology enables agentless image-level backups. Answer: VADP (vSphere Storage APIs – Data Protection). Trap: they might list ‘VAAI’ or ‘vVols’ – those are different technologies.
vSphere Replication vs Backup: They will give a scenario requiring near-real-time failover to another site and ask which technology is used. Answer: vSphere Replication. Trap: they will also list ‘Storage vMotion’ which moves VMs but does not provide DR copies.
Crash-consistent vs Application-consistent backups: You must know that crash-consistent is like pulling the power cord – it is stable but may lose in-memory data. Application-consistent backups use VMware Tools quiescing to ensure database transactions are complete. The exam loves asking: ‘Which backup type guarantees a database can be restored without inconsistency?’ Answer: application-consistent.
Full vs Incremental backups: A question might ask: ‘Which backup type reduces the backup window but requires a previous full backup to restore?’ Answer: incremental backups. Trap: they might say ‘differential’ – but VADP typically uses incremental (forward) or reverse incremental.
Backup targets: You may be asked where to store backups. Options include: local datastore (not recommended – if the host fails, backups are lost), NAS, SAN, cloud storage, or tape. The correct answer for a robust strategy is off-site or cloud storage.
Retention policies: They will test that you understand the 3-2-1 rule: three copies of the data, on two different media types, with one copy off-site.
Restore types: Full VM restore, file-level restore, Instant VM Recovery (a technology that lets you boot a VM directly from a backup file). You should know when each is appropriate.
Common trap patterns:
The question describes a scenario where the backup target is on the same datastore as the production VM. The correct answer is that this is NOT a valid backup strategy because a single hardware failure destroys both.
The question says ‘the backup must be completed within 2 hours, but the VM has 500 GB of data.’ The correct answer is to use incremental backups after a full baseline.
A question about compliance might ask for backup encryption. You must know that vSphere backup traffic can be encrypted at the storage level or by the backup software using HTTPS.
What they do NOT test:
Exact backup software brand commands (no Veeam PowerShell cmdlets).
Specific storage array commands.
Configuration of third-party products.
To prepare, create flashcards for RTO, RPO, VADP, crash-consistent, application-consistent, vSphere Replication, and the 3-2-1 rule. Practise with scenario questions: read the business requirement, identify the RTO and RPO, then choose the correct technology.
A backup is a separate copy of VM data stored on a different storage system, while a snapshot is a temporary delta file stored on the same datastore – never rely on snapshots as backups.
RTO (Recovery Time Objective) defines how quickly you must restore service; RPO (Recovery Point Objective) defines how much data loss in time is acceptable.
VADP (vSphere Storage APIs – Data Protection) is the VMware framework that allows third-party backup products to perform agentless, image-level backups.
Crash-consistent backups guarantee the data is as it would be after a power failure, but application-consistent backups guarantee database integrity through quiescing.
vSphere Replication provides near-real-time copies at a secondary site for disaster recovery, whereas traditional backups are point-in-time copies for long-term retention.
Always follow the 3-2-1 rule: three copies of your data, on two different media types, with one copy stored off-site.
A full restore test must be performed at least quarterly – success status in the backup log does not prove recoverability.
These come up on the exam all the time. Here's how to tell them apart.
Full Backup
Copies the entire VM disk every time
Uses a lot of storage space and time
Restore requires only the full backup file
Incremental Backup
Copies only changed blocks since the last backup
Saves storage and speeds up the backup window
Restore requires the full backup plus all incremental files
vSphere Snapshot
Stored on the same datastore as the VM
Designed for short-term change tracking (hours/days)
Not portable – cannot restore to different hardware
vSphere Backup
Stored on separate storage (NAS, cloud, SAN)
Designed for long-term retention (weeks/months/years)
Portable – can be restored to any ESXi host
Crash-consistent Backup
Captures data as if the power was pulled
Does not quiesce the file system
May leave database transactions incomplete
Application-consistent Backup
Uses VMware Tools to quiesce (flush) disks
Ensures all pending writes are completed
Database is in a consistent state for immediate recovery
Mistake
If I enable VMware HA (High Availability), I do not need backups because the VM will restart on another host.
Correct
HA only restarts a VM that crashes – it does not protect against data corruption, accidental deletion, or ransomware. Backups are still essential.
Beginners confuse availability (keeping the VM running) with backup (having a recoverable copy). They sound similar but solve different problems.
Mistake
A VM snapshot is the same as a backup.
Correct
A snapshot is a point-in-time delta file used for short-term changes (e.g., before a patch). It saves to the same datastore as the VM, so a disk failure destroys both. Snapshots are not portable and are not designed for long-term retention.
VMware documentation uses the word 'snapshot' for its own feature, and beginners assume it functions like a backup snapshot, which it does not.
Mistake
Backups only need to be taken once a week if the data rarely changes.
Correct
Even if the data rarely changes, the RPO defines the maximum acceptable data loss. If a disaster occurs on day 6 after a weekly backup, 6 days of data are lost. The backup schedule must align with the business RPO.
People think 'frequency' is tied to 'change rate' rather than 'acceptable loss rate'.
Mistake
Testing a restore is optional if the backup software reports 'success'.
Correct
'Success' only means the copy operation completed. The backup file could be corrupt, incomplete, or not fit for the restore method. Only a full test restore proves the data is recoverable.
Users trust software status messages without verifying the actual outcome – a classic gap in real-world IT.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
No. Snapshots are designed for short-term changes (like before a software update) and are stored on the same datastore as the VM. A single disk failure destroys both the VM and its snapshots. Always use a dedicated backup product.
A full backup copies the entire VM disk each time. An incremental backup copies only the blocks that changed since the last full or incremental backup. Incremental backups are faster but require a chain of files to restore completely.
The best practice is to test full restores at least once per quarter. The VCP-DCV exam expects you to know that verification is a continuous process, not a one-time event.
It is a best practice: keep three copies of your data (production plus two backups), on two different media types (e.g., disk and cloud), with one copy stored off-site to protect against site-wide disasters.
No. Replication is for disaster recovery – it keeps a live copy at a secondary site for quick failover. But it does not protect against data corruption or accidental deletion, because the corruption is replicated too. Traditional backups with long retention are still required.
VADP stands for vSphere Storage APIs – Data Protection. It is the set of APIs that allows backup software to perform agentless backups of VMs. It is important because it eliminates the need to install backup agents inside every guest OS.
You've finished vSphere Backup and Recovery. Continue through the VCP-DCV study guide to build a complete picture of the exam.
Done with this chapter?