Cloud Storage Classes¶

Core Concepts¶

Storage classes determine pricing and access characteristics for objects in Cloud Storage. Understanding the economic trade-offs between storage cost, retrieval cost, and minimum storage duration is critical for cost-effective architecture.

Key Principle: Lower storage cost = higher retrieval cost + minimum duration penalties. Match storage class to actual access patterns.

Storage Class Comparison¶

Class	Access Frequency	Storage Cost	Retrieval Cost	Min Duration	Availability	Use Case
Standard	>1/month	Highest	None	None	Highest	Hot data
Nearline	~1/month	Medium	Low	30 days	High	Warm data
Coldline	~1/quarter	Low	Medium	90 days	High	Cool data
Archive	<1/year	Lowest	Highest	365 days	High	Cold data
Autoclass	Unknown	Automatic	Varies	None	Varies	Unpredictable

Standard Storage Class¶

Characteristics¶

Pricing (approximate, varies by location):

Storage: $0.020 per GB/month (multi-region)
Retrieval: No charge
Operations: Class A ~$0.05/10K, Class B ~$0.004/10K
Minimum duration: None
Early deletion fee: None

Availability:

Multi-region: 99.95%
Dual-region: 99.95%
Region: 99.9%

When to Use¶

✅ Appropriate for:

Frequently Accessed Data:

Website content and assets
Streaming media actively used
Mobile/web application data
Analytics data in active use
Content being actively processed

Short-Term Storage:

Temporary processing data
Build artifacts (recent builds)
Staging data for pipelines
Cache storage

Performance-Critical:

Low latency required
High throughput needed
Frequent access (>1/month)
No retrieval delay acceptable

When NOT to Use¶

❌ Inappropriate for:

Data accessed less than monthly
Long-term archives
Backup data rarely accessed
Historical data for compliance only
Aged logs and analytics

Cost Impact: Paying premium storage cost for infrequent access wastes money

Nearline Storage Class¶

Characteristics¶

Pricing:

Storage: $0.010 per GB/month (~50% of Standard)
Retrieval: $0.01 per GB retrieved
Minimum duration: 30 days
Early deletion fee: Charged for remaining days

Availability: Same as Standard

Economics¶

Break-Even Analysis:

When is Nearline cheaper than Standard?

Standard cost: $0.020/GB/month
Nearline storage: $0.010/GB/month
Nearline retrieval: $0.01/GB

Nearline is cheaper if:
Access frequency < 1 full retrieval per month
(0.010 storage + 0.01 retrieval) < 0.020

Example: 1 TB data, accessed 50% monthly

Standard: 1000 × $0.020 = $20/month
Nearline: (1000 × $0.010) + (500 × $0.01) = $15/month
Savings: $5/month (25%)

When to Use¶

✅ Appropriate for:

Monthly Access Patterns:

Monthly reports and analytics
Backup data accessed for recovery testing
Aged data referenced occasionally
Compliance data with periodic review
Media assets for seasonal campaigns

Data Retention Requirements:

Regulatory data (30+ day retention)
Application logs (older than 30 days)
Database backups (recent history)
Development artifacts (older versions)

When NOT to Use¶

❌ Inappropriate for:

Short-Term Storage (<30 days):

Temporary processing data
Short-lived cache
Data deleted within 30 days
Rapid turnover data

Why: Early deletion fees negate cost savings

Frequent Access (>1/month):

Active application data
Frequently accessed logs
Primary datasets
Live analytics data

Why: Retrieval costs exceed Standard storage cost

Early Deletion Considerations¶

Scenario: Delete object after 20 days

Charged for 30 days minimum
Remaining 10 days × storage rate
No savings over Standard for short-term data

Architecture Decision: Only use Nearline if data lives >30 days

Coldline Storage Class¶

Characteristics¶

Pricing:

Storage: $0.004 per GB/month (~20% of Standard)
Retrieval: $0.02 per GB retrieved
Minimum duration: 90 days
Early deletion fee: Charged for remaining days

Availability: Same as Standard

Economics¶

Break-Even Analysis:

Coldline is cheaper than Standard if:
Access frequency < 0.8 full retrievals per month

Coldline is cheaper than Nearline if:
Access frequency < 0.3 full retrievals per month

Example: 1 TB data, accessed 10% quarterly (3.3% monthly)

Standard: $20/month
Nearline: $10 + $3.33 = $13.33/month
Coldline: $4 + $0.67 = $4.67/month
Savings: $15.33/month (77% vs Standard)

When to Use¶

✅ Appropriate for:

Quarterly Access:

Quarterly business reports
Disaster recovery data (tested quarterly)
Audit data (occasional review)
Seasonal data archives
Aged analytics data

Compliance and Archival:

Regulatory data (90+ day retention)
Legal documents (occasional access)
Historical records
Long-term backups (recent years)

Infrequently Accessed Backups:

Database backups (older than 90 days)
VM snapshots (disaster recovery)
Application state backups
Configuration backups

When NOT to Use¶

❌ Inappropriate for:

Short-Term Storage (<90 days):

Early deletion fees expensive
No cost benefit

Frequent Access (>1/quarter):

Retrieval costs accumulate
Nearline or Standard cheaper

Performance-Critical:

Though retrieval is fast, cost implies rare access
Use Standard for performance needs

Archive Storage Class¶

Characteristics¶

Pricing:

Storage: $0.0012 per GB/month (~6% of Standard)
Retrieval: $0.05 per GB retrieved
Minimum duration: 365 days
Early deletion fee: Charged for remaining days

Availability: Same as Standard (despite name suggesting slower access)

Economics¶

Break-Even Analysis:

Archive is cheaper than Standard if:
Access frequency < 0.38 full retrievals per month

Archive is cheaper than Coldline if:
Access frequency < 0.05 full retrievals per month

Example: 1 TB data, accessed once per year (8.3% monthly)

Standard: $20/month × 12 = $240/year
Coldline: $4/month × 12 + $16.6 = $64.6/year
Archive: $1.20/month × 12 + $4.2 = $18.6/year
Savings: $221.4/year (92% vs Standard)

When to Use¶

✅ Appropriate for:

Annual or Rare Access:

Compliance archives (multi-year retention)
Historical data (rarely accessed)
Long-term backups (disaster scenarios only)
Legal archives
Scientific data preservation

Regulatory Requirements:

7-year retention (SOX, HIPAA)
Legal hold data
Audit trails
Medical records

Cold Data:

Data rarely if ever accessed
“Just in case” storage
Regulatory compliance only
No business use but cannot delete

When NOT to Use¶

❌ Inappropriate for:

Any Regular Access:

Monthly, quarterly, even semi-annual access expensive
Very high retrieval costs
Use Coldline instead

Short-Term Storage (<365 days):

Massive early deletion fees
Most expensive option for short-term

Active Archives:

“Archive” doesn’t mean slower access
Use based on access frequency, not retrieval speed
If accessed regularly, use different class

Common Misconception: “Archive is for backups”

Reality: Archive is for RARELY accessed data
Backups may be accessed frequently (use Nearline/Coldline)
Archive is for compliance, not operational recovery

Autoclass¶

Characteristics¶

How It Works:

Automatically transitions objects between classes
Based on access patterns
Starts as Standard
Moves to Nearline after 30 days without access
Moves to Coldline after 90 days without access
Moves to Archive after 365 days without access (if enabled)
Moves back to Standard when accessed

Pricing:

Storage cost based on class object is in
No retrieval fees for automatic transitions
Small management fee (~$0.0025 per 1000 objects/month)
No minimum storage duration penalties

When to Use¶

✅ Appropriate for:

Unknown Access Patterns:

New data lake
Uncertain usage patterns
Variable access across objects
Mixed workloads

Simplification:

Don’t want to manage lifecycle policies
Automated optimization preferred
Mixed access patterns in bucket
Hands-off cost optimization

Testing and Development:

Unpredictable access
Changing requirements
Experimental data

When NOT to Use¶

❌ Inappropriate for:

Known Access Patterns:

Predictable usage (set specific class)
Consistent access frequency
Better cost control with manual selection

Frequent Access:

If all data accessed frequently, use Standard
Autoclass overhead not needed

Compliance Requirements:

Need specific storage class guarantees
Regulatory requirements for storage type
Audit requirements

Cost Optimization:

Autoclass convenient but may not be cheapest
Manual lifecycle policies can be more cost-effective
Pay for management overhead

Trade-offs¶

Benefits:

Automatic optimization
No lifecycle policy management
Adapts to changing patterns
Simple to implement

Drawbacks:

Management fee
Less control
May not be optimal for all objects
Harder to predict costs

Decision Framework¶

Access Pattern Analysis¶

Questions to Answer:

How often will data be accessed?
Daily/weekly → Standard
Monthly → Nearline
Quarterly → Coldline
Yearly/rare → Archive
How long will data be stored?
<30 days → Standard only
30-90 days → Standard or Nearline
90-365 days → Standard, Nearline, or Coldline
365 days → Any class
What percentage retrieved when accessed?
Full dataset → Factor full retrieval cost
Partial → Lower effective retrieval cost
Is access pattern predictable?
Yes → Choose specific class
No → Consider Autoclass
Are there minimum storage requirements?
Short-term data → Beware early deletion fees
Long-term data → All classes viable

Cost Calculation Formula¶

Total Monthly Cost = Storage Cost + Retrieval Cost + Operation Cost

Standard:    (GB × $0.020) + (0 retrieval) + ops
Nearline:    (GB × $0.010) + (GB_retrieved × $0.01) + ops
Coldline:    (GB × $0.004) + (GB_retrieved × $0.02) + ops
Archive:     (GB × $0.0012) + (GB_retrieved × $0.05) + ops

Example: 10 TB stored, 20% accessed monthly

Standard: 10,000 × 0.020 = $200
Nearline: (10,000 × 0.010) + (2,000 × 0.01) = $120
Coldline: (10,000 × 0.004) + (2,000 × 0.02) = $80
Archive: (10,000 × 0.0012) + (2,000 × 0.05) = $112

Best choice: Coldline ($80/month)

Multi-Class Bucket Strategy¶

Architectural Pattern¶

Single Bucket, Multiple Classes:

Objects can have different storage classes in same bucket
Bucket has default class for new objects
Individual objects can override
Lifecycle policies transition between classes

Use Case: Data aging pattern

New objects → Standard (default)
After 30 days → Nearline (lifecycle policy)
After 90 days → Coldline (lifecycle policy)
After 365 days → Archive (lifecycle policy)
After 7 years → Delete (compliance)

Common Mistakes¶

Mistake 1: Using Archive for Backups¶

Problem: Backups often accessed for testing and recovery

Solution: Use Nearline or Coldline based on access frequency

Mistake 2: Ignoring Early Deletion Fees¶

Problem: Delete Nearline object after 20 days, pay for 30 days

Solution: Use Standard for short-lived data

Mistake 3: Overusing Standard¶

Problem: Paying premium for data accessed monthly

Solution: Analyze access patterns, use Nearline for monthly access

Mistake 4: Underestimating Retrieval Costs¶

Problem: Archive looks cheap but retrieval costs accumulate

Solution: Calculate total cost including retrieval frequency

Mistake 5: Autoclass for Everything¶

Problem: Management fees and less control

Solution: Use Autoclass only when access patterns unknown

Exam Focus Areas¶

Storage Class Selection¶

Economic trade-offs (storage vs retrieval vs minimum duration)
Access pattern matching
Break-even analysis
Early deletion fee scenarios

Cost Optimization¶

Lifecycle policies to transition classes
Appropriate class for use case
Total cost calculation (not just storage)
Multi-class bucket strategies

Architecture Patterns¶

Data aging strategies
Backup and archive distinction
Compliance requirements
Performance vs cost trade-offs

Common Scenarios¶

Hot/warm/cold data patterns
Compliance and retention
Disaster recovery storage
Data lake storage classes

Decision Making¶

When to use each class
When NOT to use each class
Autoclass vs manual selection
Multi-class bucket design