Data protection options

This document compares the options available on Compute Engine to back up and replicate the following Compute Engine resources:

  • Persistent Disk and Google Cloud Hyperdisk volumes
  • Virtual machine (VM) instances
  • Workloads running in Compute Engine and on-premises

Options

Feature Summary Use case Optimized for
Backup and DR Service

Managed backup and disaster recovery (DR) service for centralized data protection of VMs and other workloads running in Google Cloud and on-premises.
  • Centralized, application-consistent data protection.
  • Automated backups directed by a backup plan for retention of short term (intraday, daily, weekly) and long term (monthly, yearly) backups.
  • Cross-project and cross-region recoveries to enable disaster recovery.
  • Full and incremental backups using changed-block tracking.
 Centralized backup management of VMs and databases across projects.
Standard snapshots
  • Captures the state of your disk at a particular point in time.
  • Stored as differential copies for better performance and space efficiency.
 Long term data backup. Long retention backup and geo-redundancy.
Archive snapshots
  • Offers all the same benefits as standard snapshots, but at a lower-cost.
  • Suited for use cases related to compliance, audit, and long-term cold storage.
 Long term data backup that is rarely accessed but must be retained for several months or years. Long retention backup and geo-redundancy.
Instant snapshots
  • Captures the state of your disk at a particular point in time.
  • Each instant snapshot is stored as a differential copy of the disk for more efficient storage and better performance.
 Quick local disk backup that enables rapid data restoration in case of application failure or user error.
  • Rapid data restoration after application corruption, user error, or failed upgrades.
  • Low RTO and RPO.
Machine images Stores all the configuration, metadata, permissions, and data from one or more disks required to create a virtual machine (VM) instance.
  • Long term VM backup and restore.
  • Instance cloning and replication.
 VM consistency at the I/0 operation level or crash level.
Regional persistent disk Replicates data synchronously across two zones in the same region.
  • High availability.
  • Not designed for data backup.
 RTO: less than 1 minute
RPO: 0
Disk clones Creates a live, attachable, fully provisioned disk with data from a source disk.
  • Quickly bringing up staging environments from production.
  • Replication for backup verification or export offloading.
  • Not designed for disaster recovery.
 Copy data management.
Images Contains the set of programs and files required to boot an operating system on a VM instance.
  • VM import and fast disk creation.
  • Quickly creating many VMs.
 Rapid disk creation.

Best practices

Avoid temporary standard snapshots

To immediately create a copy of a disk in the same zone for verification or export, use disk clones or instant snapshots instead of standard snapshots. Compared to disk clones and instant snapshots, standard snapshots have longer copy times for upload and download.

Schedule hourly standard snapshots for backup and disaster recovery

Schedule hourly standard snapshots. If you require daily snapshots, consider scheduling snapshots every 6 hours.

Use images for fast and frequent disk creation across regions

To create many disks from a single data source, use images instead of snapshots. Because Compute Engine performs local caching in target zones, disk creation from images is faster than disk creation from snapshots.

To create backups of all disks attached to a VM instance, use machine images

Use machine images to create backups of all disks that are attached to a VM. A machine image can be used to backup multiple disks at a time to help ensure that the data captured in the machine image is consistent across all disks. A persistent disk snapshot can only back up a single disk at a time. For more information, see When to use machine images.

What's next