While preparing for the AWS SAA-C03, many candidates get confused by hybrid storage migration strategies. In the real world, this is fundamentally a decision about latency requirements vs. storage economics. Let’s drill into a simulated scenario.
The Scenario #
GlobalMedia Productions operates an on-premises Windows SMB file server that stores large video editing project files. Their creative teams heavily access newly created assets—approximately 85% of file operations occur within the first 7 days after creation. After this initial period, files are rarely accessed but must be retained for compliance and occasional reference for up to 7 years.
The company’s storage infrastructure is reaching 92% capacity, and procurement cycles for additional SAN hardware take 6-8 weeks. The IT Director needs a solution that:
- Immediately increases available storage capacity
- Preserves sub-50ms latency for recently created files
- Reduces long-term storage costs through automated tiering
- Requires minimal changes to existing user workflows (users currently access files via SMB shares)
Key Requirements #
Extend storage capacity immediately while maintaining low-latency access to recently created files, and implement automated lifecycle management to control future storage growth and costs.
The Options #
- A) Use AWS DataSync to replicate files older than 7 days from the SMB file server to AWS.
- B) Deploy an Amazon S3 File Gateway to extend storage capacity, configure S3 lifecycle policies to transition objects to S3 Glacier Deep Archive after 7 days.
- C) Create an Amazon FSx for Windows File Server file system to extend the company’s storage capacity.
- D) Install Amazon S3 client utilities on each user workstation to access S3 directly, create S3 lifecycle policies to transition data to S3 Glacier Flexible Retrieval after 7 days.
Correct Answer #
Option B.
The Architect’s Analysis #
Correct Answer #
Option B - Deploy Amazon S3 File Gateway with S3 lifecycle policies.
Step-by-Step Winning Logic #
S3 File Gateway provides the optimal latency-cost trade-off for this hybrid storage scenario:
1. Transparent SMB Integration
- Users continue accessing files via familiar SMB protocol (no workflow changes)
- Gateway presents S3 storage as a standard Windows file share
- No client-side software installation or training required
2. Intelligent Caching Architecture
- Gateway appliance caches frequently accessed files locally (your “hot” 7-day window)
- Sub-50ms latency for cached data (meets the low-latency requirement)
- Asynchronous upload to S3 in the background
3. Automated Lifecycle Economics
- S3 lifecycle policies automatically transition aging data through storage tiers
- Days 1-7: S3 Standard (~$23/TB/month) with local cache
- Day 8+: S3 Glacier Deep Archive (~$1/TB/month) = 95% cost reduction
- No manual intervention required—meets “avoid future storage problems” requirement
4. Immediate Capacity Expansion
- Gateway deployed as VM or hardware appliance in days, not weeks
- Effectively unlimited S3 backend storage (no more capacity planning cycles)
The Traps (Distractor Analysis) #
Why not Option A (AWS DataSync)?
- DataSync is a one-way migration/sync tool, not a transparent storage extension
- Does NOT provide SMB access to data after migration to AWS
- Users would lose access to migrated files through existing workflows
- Requires building an entirely new access pattern (violates minimal workflow change requirement)
- Cost model: Adds DataSync transfer fees (~$0.0125/GB) without solving the access latency problem
Why not Option C (Amazon FSx for Windows File Server)?
- FSx provides excellent SMB performance but at much higher cost (~$0.13-0.65/GB/month depending on throughput)
- No automated lifecycle management to cheaper storage tiers
- 100% of data remains in expensive high-performance storage
- For a 100TB dataset: FSx = ~$13,000-65,000/month vs. File Gateway with lifecycle = ~$2,300/month initially, dropping to ~$100/month for archived data
- Doesn’t address “avoid future storage problems”—just moves the cost burden to AWS
Why not Option D (S3 client tools on workstations)?
- Catastrophic user experience change: completely abandons SMB workflow
- Users must learn new tools and access patterns
- Applications expecting SMB paths will break
- S3 API calls have higher latency than SMB for small file operations
- Glacier Flexible Retrieval has 1-5 minute retrieval time (violates low-latency requirement for any archived file access)
- No local caching mechanism
The Architect Blueprint #
Diagram Note: Users access S3 storage transparently via SMB protocol through the File Gateway, which caches hot data locally and automatically tiers cold data to Glacier Deep Archive, creating a cost-optimized hybrid storage architecture.
The Decision Matrix #
| Option | Est. Complexity | Est. Monthly Cost (100TB Dataset) | Pros | Cons |
|---|---|---|---|---|
| A: DataSync | Medium | $1,250 initial transfer + S3 storage (~$2,300/month) = $3,550 first month, then $2,300 | • Simple one-way migration • Good for backup scenarios |
• Breaks SMB access • No transparent storage extension • Doesn’t solve latency requirement |
| B: S3 File Gateway ✅ | Low-Medium | Month 1-2: ~$2,400 Month 12: ~$350 (80% tiered to Glacier) Long-term: ~$150/month |
• Transparent SMB access • Local cache for low latency • Automated lifecycle = 95% cost reduction • Unlimited scalability |
• Initial gateway setup • Cache size planning needed • Retrieval latency for archived data |
| C: FSx for Windows | Low | $13,000-$65,000/month (depending on throughput tier) | • Native Windows experience • High performance • Active Directory integration |
• 10-40x more expensive • No automated tiering to cold storage • Doesn’t address future growth economics |
| D: S3 Direct Access | High | Month 1-2: ~$2,300 Month 12: ~$300 (with lifecycle) |
• Lowest storage cost potential • Direct cloud integration |
• Requires user retraining • Application compatibility issues • Glacier retrieval = 1-5 min latency • No local caching |
Cost Calculation Notes:
- S3 Standard: $0.023/GB/month = $23/TB
- Glacier Deep Archive: $0.00099/GB/month = $1/TB
- FSx Windows (50 MB/s): $0.13/GB/month = $130/TB
- DataSync: $0.0125/GB one-time transfer
Real-World Practitioner Insight #
Exam Rule #
For the AWS SAA-C03 exam, when you see:
- “SMB file server” + “low latency for recent files” + “lifecycle management” → Choose S3 File Gateway with lifecycle policies
- “Transparent access” + “minimal user workflow changes” → Storage Gateway family (not DataSync or direct S3 access)
- “7 days hot, then cold” → Perfect use case for S3 lifecycle transitions
Real World #
In production environments, we’d enhance this architecture with:
1. Hybrid Cache Sizing
- Calculate cache size based on actual working set:
(Daily new data × 7 days) + 20% buffer - For GlobalMedia’s scenario: If they create 2TB/day, cache appliance needs ~17TB local storage
- Consider multiple gateway appliances for high-availability
2. Multi-Tiered Lifecycle
- Days 0-7: S3 Standard (hot cache)
- Days 8-90: S3 Intelligent-Tiering (automatically optimizes)
- Days 91-365: S3 Glacier Flexible Retrieval (3-5 hour retrieval acceptable)
- Day 366+: S3 Glacier Deep Archive (12-hour retrieval for compliance-only access)
- This approach saves an additional 15-25% vs. direct Standard→Deep Archive transition
3. Bandwidth Considerations
- File Gateway requires adequate bandwidth: Estimate
(Daily change rate × 1.5)for comfortable asynchronous upload - For 2TB/day workload: Minimum 500 Mbps dedicated connection recommended
- Consider AWS Direct Connect if on-premises link is constrained
4. Monitoring & Alerting
- CloudWatch metrics:
CachePercentDirty(data not yet uploaded) CacheHitPercent(efficiency of cache sizing)- Set alarms when cache hit rate drops below 85% (indicates undersized cache)
5. Disaster Recovery Enhancement
- S3 Cross-Region Replication for business-critical video assets
- Versioning enabled to protect against accidental deletions
- MFA Delete for compliance environments
The Exam Simplification: The exam question intentionally omits these nuances to test your foundational understanding. In real projects, we’d also evaluate:
- Existing MPLS/Direct Connect infrastructure costs
- Whether the company needs multi-site access (favors FSx for Windows with Multi-AZ)
- Actual retrieval SLAs for archived content (might influence Glacier tier selection)
- Integration with existing backup solutions (Veeam, Commvault, etc.)