Skip to main content

Overview

Xloud Compute provides two complementary data protection mechanisms for virtual machine instances. Instance snapshots capture the full disk state as a reusable image, and volume backups archive attached block storage to object storage. Together they give you point-in-time recovery, pre-change safety nets, and a reliable path for cloning and migrating workloads.
Prerequisites
  • A running or shutoff instance
  • Sufficient quota in the Image Service (for snapshots) and Object Storage (for backups)
  • Volumes attached to the instance if you want volume-level backups

Instance Snapshots

An instance snapshot saves the entire disk state of a running or shutoff VM as an image in the Xloud Image Service. You can boot new instances from the snapshot or use it as a restore point.

Open the instance actions menu

Navigate to Compute → Instances, find your instance, and click Create Snapshot from the actions dropdown.

Name the snapshot

Enter a descriptive name. The snapshot is stored in the Image Service under your project.
Use a naming convention like <instance-name>-YYYYMMDD-pre-<change> so snapshots are easy to identify and sort.

Wait for completion

The instance status changes to IMAGE_SNAPSHOT while the snapshot is being captured. For large disks this may take several minutes.
The snapshot appears under Compute → Images with status Active when complete.

Restore from Snapshot

Boot a new instance from any saved snapshot to restore the workload or create a clone:
Navigate to Compute → Images, find the snapshot, and click Launch Instance. The instance boots with the exact disk state from the snapshot.

Volume Backups

For instances with attached persistent volumes, use block storage backups to archive volume data to object storage independently of the instance lifecycle.
Navigate to Block Storage → Volumes, select a volume, and click Create Backup from the actions menu. Enter a name and optionally select an existing snapshot as the source.
Volume backups are stored in Object Storage and persist independently of the source volume. They can be restored to any volume in the same project.

Snapshot Management

Use the CLI in a scheduled task to create regular snapshots:
Weekly snapshot script
#!/bin/bash
INSTANCE="web-server-01"
DATE=$(date +%Y%m%d)
source ~/openrc.sh

openstack server image create \
  --name "${INSTANCE}-weekly-${DATE}" \
  "${INSTANCE}"

# Keep only last 4 weekly snapshots
openstack image list \
  --property image_type=snapshot \
  --format value -c ID -c Name | \
  grep "${INSTANCE}-weekly" | \
  sort -k2 | head -n -4 | \
  awk '{print $1}' | \
  xargs -r openstack image delete
Snapshots are private by default. To share with another project:
Make snapshot shared
openstack image set --shared my-instance-snap-20250315

openstack image add project \
  my-instance-snap-20250315 \
  <target-project-id>
After validating a snapshot as a stable base, set its visibility to public (admin only) or promote it as a project-shared image:
Promote to shared project image
openstack image set \
  --shared \
  --property description="Hardened Ubuntu 22.04 base" \
  my-instance-snap-20250315

Snapshot vs Backup — Comparison

FeatureInstance SnapshotVolume Backup
What is capturedFull instance disk (root + ephemeral)Single attached volume
Storage locationImage ServiceObject Storage
Used toBoot new instances, clone, restoreRestore volume data
BootableYesNo (restore to volume first)
Incremental supportNo (always full)Yes (incremental after first)
Cross-project restoreYes (with share)No (project-scoped)
Snapshots capture disk state at a point in time. For application-consistent backups of databases or stateful services, quiesce writes (or use application-level dumps) before creating a snapshot.

Application-Consistent Snapshots

Xloud-Developed — This capability is developed by Xloud and ships with XAVS / XPCI.
Standard snapshots are crash-consistent — they capture whatever is on disk at that instant. Application-consistent snapshots go further by coordinating with the guest operating system to flush buffers and freeze filesystems before the snapshot is taken.
FeatureDescription
Filesystem freeze (Linux)The XAVS Guest Agent issues fsfreeze --freeze to quiesce all mounted filesystems before the snapshot, then thaws automatically after
VSS integration (Windows)On Windows guests, the agent triggers a Volume Shadow Copy Service (VSS) snapshot for application-level consistency
Strict quiesce modeSet os_require_quiesce=yes on the image metadata to fail the snapshot if the XAVS Guest Agent is not responding — prevents silently falling back to crash-consistent snapshots
Custom pre/post hooksConfigure pre-snapshot and post-snapshot scripts inside the guest (e.g., pg_start_backup / pg_stop_backup for PostgreSQL) for database-level consistency
Dashboard status indicatorsThe Dashboard displays live XAVS Guest Agent connectivity status and consistency badges (application-consistent vs. crash-consistent) on each snapshot
Install the XAVS Guest Agent on all production instances. Without the agent, snapshots fall back to crash-consistent mode, which may require filesystem repair (fsck) or database recovery on restore. See File-Level Restore for browsing snapshot contents without a full restore.

Agentless File-Level Restore

Xloud-Developed — This capability is developed by Xloud and ships with XAVS / XPCI.
Xloud provides agentless file-level restore directly from instance snapshots — browse directories, preview files, and download individual files or multi-file ZIP archives without restoring the entire virtual machine. This is a major competitive differentiator: no in-VM agent is required, and the feature works even on snapshots of decommissioned or deleted instances.
CapabilityDescription
Partition detectionAutomatically detects all partitions within the snapshot image (Linux ext4/xfs/btrfs, Windows NTFS)
Directory navigationBrowse the full filesystem tree from the Dashboard — navigate folders, view file sizes and timestamps
Single file downloadDownload any individual file directly from the snapshot to your local machine
Multi-file ZIP archiveSelect multiple files or directories and download them as a single ZIP archive
No agent dependencyWorks without any software inside the guest — operates at the block device layer on the storage backend
Access file-level restore from the Compute / Images panel in the Dashboard. Select any snapshot and click Browse Files to open the file browser.
File-level restore is ideal for recovering configuration files, database dumps, or application logs from snapshots without the time and resource cost of booting a full restored instance.

Next Steps

Image Service

Manage, share, and convert snapshot images in the Xloud Image catalog

Block Storage Snapshots

Volume-level snapshots for persistent block storage attached to instances

Disaster Recovery

Automated failover and recovery for mission-critical instances

Instance Launch

Boot new instances from snapshots or base images