Skip to content

Machine Images and Disk Images

Core Concepts

Images capture VM or disk state for replication, backup, and standardization. Understanding the differences between machine images, disk images, and snapshots is critical for architecture decisions.

Machine Images vs Disk Images vs Snapshots

Feature Machine Image Disk Image Snapshot
Scope Entire VM Single disk Single disk
Configuration Yes (all settings) No No
Multiple Disks Yes (all attached) No No
Storage Method Full copy Full copy Incremental
Cost Highest Medium Lowest (incremental)
Creation Time Slowest Medium Fast (after first)
Use Case VM cloning, DR OS templates Regular backups
Frequency Weekly/monthly One-time Daily/hourly

Machine Images

Architecture Purpose

Complete VM Backup:

  • All disk data preserved
  • VM configuration captured (machine type, labels, network settings)
  • Metadata and startup scripts included
  • Service account configuration retained

When to Use:

  • Complete VM backup and restore
  • VM migration between projects/regions
  • Disaster recovery (weekly/monthly full backups)
  • Creating identical VMs across environments
  • Compliance snapshots (full system state)

When Not to Use:

  • Regular incremental backups (use snapshots instead)
  • Single disk backup (use snapshots)
  • OS templates (use disk images)
  • Frequent backups (too expensive)

Design Patterns

DR Strategy:

  • Weekly machine images for complete recovery
  • Daily snapshots for incremental backups
  • Store in multi-regional location
  • Test restoration quarterly

VM Migration:

  • Create machine image in source project
  • Share with target project via IAM
  • Restore in target project/region
  • Update network and external integrations

Environment Replication:

  • Production machine image
  • Restore as staging environment
  • Modify labels and metadata
  • Separate network configuration

Disk Images

Architecture Purpose

OS and Software Templates:

  • Base operating system
  • Pre-installed software
  • Security hardening applied
  • Configuration management tools
  • Company standards enforced

When to Use:

  • Creating standard VM configurations
  • Instance templates for MIGs
  • Golden images for teams
  • Consistent deployments
  • OS patching strategy (update image, redeploy VMs)

When Not to Use:

  • Backup purposes (use snapshots)
  • Capturing running VM state (use machine images)
  • Incremental updates (use configuration management)

Image Families

Concept: Logical grouping where latest image used automatically

Benefits:

  • Version control for images
  • Automatic use of latest
  • Easy rollback to previous
  • Staged rollouts

Pattern:

Image Family: "web-server"
├── web-server-v1 (deprecated)
├── web-server-v2 (obsolete)
└── web-server-v3 (active, automatically used)

Architecture Decision: Always use image families in instance templates, never specific image names

Public Images

Use Cases:

  • Quick testing/development
  • Standard OS deployments
  • Community-maintained images
  • Starting point for custom images

Considerations:

  • No customization
  • Google or community maintained
  • May not meet security standards
  • Use as base for custom images

Pattern: Public image → Customize → Save as custom image in family

Image Lifecycle Management

Deprecation States

States:

  • ACTIVE: Available for use
  • DEPRECATED: Available but not recommended, shows warning
  • OBSOLETE: Available but strongly discouraged, shows error
  • DELETED: No longer available

Strategy:

  1. Release new image (ACTIVE)
  2. Deprecate old image (after testing period)
  3. Mark obsolete (after migration period)
  4. Delete (after retention period)

Architecture Pattern: Gradual deprecation allows staged migrations

Cross-Project Sharing

Use Cases

Organization Pattern:

  • Central image repository project
  • Shared across all projects
  • Consistent base images
  • Centralized maintenance

Security Pattern:

  • Share via IAM (not public)
  • Principle of least privilege
  • Service account access
  • Audit image usage

Multi-Project Architecture:

  • Host project: Maintains images
  • Service projects: Consume images
  • Reduces duplication
  • Single source of truth

Image Creation Strategies

Build Pipeline Pattern

Automated Image Creation:

  1. Base public image
  2. Packer or Cloud Build
  3. Apply configurations
  4. Security hardening
  5. Testing and validation
  6. Promote to image family
  7. Deprecate old version

Benefits:

  • Consistent, repeatable process
  • Version controlled configurations
  • Automated testing
  • Audit trail

Golden Image Approach

Single Source of Truth:

  • Create “perfect” VM
  • Install and configure software
  • Harden and document
  • Create image from disk
  • Distribute to teams

Considerations:

  • Manual process initially
  • Requires documentation
  • Update process needed
  • Drift over time

Recommendation: Automate with Cloud Build

Cost Considerations

Storage Costs

Machine Images:

  • Charged for all disk data
  • Multiple disks = higher cost
  • Full copy each time
  • Expensive for frequent backups

Disk Images:

  • Single disk storage
  • Full copy
  • Long-term storage
  • Moderate cost

Snapshots:

  • Incremental storage
  • First snapshot = full copy
  • Subsequent = only changes
  • Most cost-effective for frequent backups

Optimization Strategies

Hybrid Approach:

  • Machine images: Weekly/monthly
  • Snapshots: Daily
  • Balance cost vs recovery options

Retention Policies:

  • Keep recent machine images (30 days)
  • Archive older snapshots
  • Delete obsolete images
  • Regular cleanup

Compliance and Security

Image Hardening

Security Baselines:

  • CIS benchmarks applied
  • Unnecessary services disabled
  • Security patches current
  • Logging configured
  • Security agents installed

Compliance Requirements:

  • Approved software list
  • Vulnerability scanning
  • Configuration standards
  • Audit logging

Scanning and Validation

Container Analysis:

  • Scan images for vulnerabilities
  • Block deployment if critical issues
  • Continuous monitoring
  • Compliance reporting

Testing:

  • Automated security tests
  • Vulnerability assessments
  • Penetration testing
  • Regular updates

Disaster Recovery

Multi-Region Strategy

Pattern:

  • Machine images in multi-regional storage
  • Snapshots in multiple regions
  • Regular DR testing
  • Documented procedures

Considerations:

  • Storage location affects cost
  • Restore time varies by location
  • Network egress charges
  • Compliance requirements

Recovery Time Objectives

Machine Images:

  • RTO: 15-30 minutes
  • Complete VM restore
  • All configuration preserved
  • Fastest recovery option

Snapshots:

  • RTO: 30-60 minutes
  • Disk restoration
  • VM recreation required
  • Configuration must be reapplied

Instance Templates

Relationship with Images

Instance Template Contains:

  • Reference to image (family preferred)
  • Machine type specification
  • Network configuration
  • Metadata and labels
  • Startup scripts
  • NOT the actual disk data

Architecture Pattern:

Disk Image (OS + Software)
    ↓
Image Family (versioning)
    ↓
Instance Template (configuration)
    ↓
Managed Instance Group (scaling)

Versioning Strategy

Pattern:

  • Update image in family
  • Instance template references family
  • MIG rolling update to new version
  • Automated, consistent deployments

Benefits:

  • Decoupling of image and configuration
  • Easy rollbacks
  • Gradual rollouts
  • Version control

Exam Focus Areas

Decision Criteria

  • Machine image vs disk image vs snapshot trade-offs
  • When to use each type
  • Cost comparison and optimization
  • Storage location decisions

Architecture Patterns

  • Image pipeline automation
  • Golden image management
  • Cross-project sharing
  • Image family versioning

DR and Backup

  • Hybrid backup strategies (images + snapshots)
  • Multi-region DR with images
  • RTO/RPO considerations
  • Testing and validation

Security

  • Image hardening strategies
  • Vulnerability scanning
  • Compliance baselines
  • Access control (IAM)

Operations

  • Image lifecycle management
  • Deprecation strategies
  • Instance template integration
  • Automated updates