When working with a virtual server, you can create a snapshot that captures the current state of the system before an update, configuration change, or installation of new software. In the control panel, it often appears as another type of backup: if the system can be restored, then the data is protected. The error becomes noticeable when the problem goes beyond failed changes: the storage is corrupted, the virtual machine disappears, or an attacker gains access to the infrastructure.
A snapshot and a backup indeed allow for data recovery. However, they are designed for different situations, have different retention periods, and depend on the underlying system in different ways.
A snapshot captures the state of the system, a backup creates a fallback
Snapshot, or image, captures the state of the disk or virtual machine at a specific moment. It is convenient to create before updating the operating system, changing the web server configuration, or deploying a new version of the application. If the site stops working after the changes, the server can quickly be reverted to its previous state.
A snapshot does not copy all the data of the virtual server to a separate storage. It captures the state of the virtual machine at a specific moment, and subsequent changes are recorded separately. If you need to go back, the system uses the primary disk and the snapshot data to restore the previous state.
That is why a snapshot is created quickly and usually takes up less space than a full backup. At the same time, it remains linked to the primary disk and the infrastructure on which the virtual server operates. If the source data or the storage itself is lost, a single snapshot may not be enough for recovery.
Backup is designed for longer storage and a broader set of events. It can contain individual files, a database, configurations, a disk image, or the entire system. For backups, a schedule, retention depth, encryption rules, and copy locations are set.
The difference is not that a backup is necessarily large and slow, while a snapshot is always created in seconds. Backups can also be incremental. The main question is phrased differently: will the recovery point remain accessible if the primary system is lost.
Read also: How to choose an operating system for a VPS server
Why a snapshot alone is not enough
Before updating PHP, the administrator created a snapshot. The new version turned out to be incompatible with the site, so the server was reverted to its previous state. In this case, the snapshot works well: recovery is quick, and there is no point in keeping it for months.
Another situation: an employee deleted a directory with documents, but the error was noticed five days later. The short-term snapshot may have already been deleted. Here, a series of backups with sufficient retention depth is needed.
There is also a more dangerous scenario. An attacker gained administrative access and deleted the server along with the snapshots. If all points were in the same account, copies technically existed, but they did not survive the crash.
Snapshots should also not be left unchecked. In systems where they create a chain of dependent disks, snapshots gradually take up more space and can affect performance. VMware recommends not keeping a single snapshot for more than 72 hours and limiting the chain to two or three snapshots. This is a recommendation for a specific platform, but it clearly shows the purpose of the technology: short-term insurance before changes, not archiving.
IMPORTANT TO KNOW! Difficulties can arise if a snapshot is created during active system operation. For example, an online store is currently recording a new order in the database. Part of the information has already been saved, while part is still being processed. If the state of the server is captured at this moment, the operation may end up in the snapshot incomplete.
After reverting to such a snapshot, the server will most likely start, but the last changes may be restored incorrectly. Therefore, before creating a snapshot of critical systems, data recording is temporarily paused, or tools are used that first complete ongoing operations.
Can a snapshot be created on a dedicated server
On a dedicated server, there is usually no ready button to create a snapshot, as in the VPS/VDS control panel. The client receives a separate physical machine, and the ability to store its intermediate states depends on the installed system and the way disks are organized.
Snapshots can be configured at the file system or storage level, for example, using ZFS or LVM. Another option is to install a virtualization platform on the physical server, divide its resources among several virtual machines, and create snapshots of each of them. This approach is used, in particular, when several sites, applications, or work environments need to be isolated on a single dedicated server.
It is better to configure this capability before launching the project, as it requires properly organizing the disks and ensuring free space. If the snapshot is stored on the same disk array, it will not protect against physical drive failures or the loss of the entire server. Therefore, on a dedicated machine, snapshots are also primarily used for quick recovery after failed changes, while backups are stored separately.
Read also: Optimizing the educational process with Windows VPS
How to combine snapshots and backups
The working scheme does not force you to choose only one tool. It distributes tasks between them:
-
a snapshot is created before an update, reconfiguration, or deployment and deleted after system verification;
-
backups are formed automatically and have multiple recovery points;
-
at least one copy is stored separately from the main server and is not deleted with the same credentials;
-
for databases, consistent copying is configured, not a mechanical snapshot of the active disk;
-
recovery is periodically tested on a test machine, as completed copying does not guarantee data integrity.
For critical information, it is advisable to follow the 3-2-1 rule: have three copies of data, use two different types of storage, and keep one copy off the main infrastructure. This is not a ready scheme for every project, but it removes dependency when the server, snapshots, and backups can be lost due to a single event.
A snapshot reduces the time to recover after failed changes. A backup protects the history of data and allows recovery after a more serious crash. Reliability comes when these tools work as different levels of protection.