Task Statement 3.1: Determine high-performing and/or scalable storage solutions.
📘AWS Certified Solutions Architect – (SAA-C03)
1. What Does “Performance” Mean in Storage?
In AWS, storage performance mainly depends on three key factors:
1. Latency
- How fast data is retrieved (delay)
- Measured in milliseconds (ms)
- Lower latency = faster response
2. Throughput
- Amount of data transferred per second
- Measured in MB/s or GB/s
- Important for large data transfers
3. IOPS (Input/Output Operations Per Second)
- Number of read/write operations per second
- Important for databases and transactional systems
2. Choosing the Right Storage Based on Performance Needs
A. Object Storage – Amazon S3
Best for:
- Large-scale storage
- Static content
- Backup and archive
Performance Characteristics:
- Virtually unlimited scalability
- High throughput
- Moderate latency (not ultra-low)
Performance Optimization:
- Use multipart upload for large files
- Use S3 Transfer Acceleration for global uploads
- Use S3 Intelligent-Tiering for cost-performance balance
Exam Tip:
- If question mentions massive scalability + high throughput → choose S3
B. Block Storage – Amazon EBS
Best for:
- Databases
- Applications running on EC2
Volume Types and Performance:
1. General Purpose SSD (gp3 / gp2)
- Balanced performance
- Up to 16,000 IOPS (gp3)
2. Provisioned IOPS SSD (io1 / io2)
- High-performance workloads
- Up to 64,000+ IOPS
- Low latency
3. Throughput Optimized HDD (st1)
- High throughput, low IOPS
- Good for large sequential workloads
4. Cold HDD (sc1)
- Lowest cost
- Very low performance
Performance Optimization in EBS:
1. Choose Right Volume Type
- Database → io2
- General apps → gp3
2. Adjust IOPS and Throughput (gp3)
- You can independently configure:
- IOPS
- Throughput
3. Use EBS-Optimized EC2 Instances
- Dedicated bandwidth for storage
4. RAID Configuration
- RAID 0 → increases performance (striping)
- RAID 1 → increases durability
Exam Tip:
- High IOPS + low latency → io2
- Balanced + cost-effective → gp3
C. File Storage – Amazon EFS
Best for:
- Shared file systems
- Multiple EC2 instances
Performance Modes:
1. General Purpose Mode
- Low latency
- Best for most applications
2. Max I/O Mode
- High throughput
- Slightly higher latency
Throughput Modes:
1. Bursting Throughput
- Performance depends on storage size
2. Provisioned Throughput
- Set required throughput manually
3. Elastic Throughput
- Automatically adjusts
Performance Optimization:
- Use Provisioned Throughput for consistent workloads
- Use Elastic Throughput for unpredictable workloads
Exam Tip:
- Shared storage + scalable → EFS
- If high throughput needed → choose Max I/O mode
D. High-Performance File Storage – Amazon FSx
Types:
1. FSx for Windows File Server
- Windows-based workloads
2. FSx for Lustre
- High-performance computing (HPC)
FSx for Lustre Performance:
- Extremely high throughput
- Sub-millisecond latency
- Integrates with S3
Exam Tip:
- HPC / big data / ML workloads → FSx for Lustre
3. Matching Storage to Workload Type
| Workload Type | Best Storage | Key Requirement |
|---|---|---|
| Database | EBS (io2) | High IOPS, low latency |
| Web servers | EBS (gp3) | Balanced performance |
| Shared file system | EFS | Multi-instance access |
| Big data analytics | FSx for Lustre | High throughput |
| Backup/archive | S3 | Scalability |
4. Important Configuration Techniques for Performance
A. Caching
- Use Amazon CloudFront or caching layers
- Reduces direct storage access
- Improves latency
B. Data Partitioning (Sharding)
- Split data across multiple volumes
- Improves performance
C. Parallelism
- Read/write data in parallel
- Used in:
- S3 multipart upload
- RAID setups
D. Compression
- Reduces data size
- Improves throughput
E. Network Optimization
- Use enhanced networking
- Use placement groups for low latency
5. Performance vs Cost Trade-Off
| Storage | Performance | Cost |
|---|---|---|
| io2 | Highest | Expensive |
| gp3 | Balanced | Medium |
| st1 | High throughput | Low |
| sc1 | Lowest | Cheapest |
Exam Tip:
- AWS questions often ask:
- “Most cost-effective solution with performance”
- Not always highest performance
6. Key Exam Scenarios to Remember
Scenario 1:
- Need ultra-fast database
→ Use EBS io2 with high IOPS
Scenario 2:
- Shared file system across EC2
→ Use EFS
Scenario 3:
- High-performance analytics
→ Use FSx for Lustre
Scenario 4:
- Massive data storage with scaling
→ Use S3
Scenario 5:
- Predictable throughput needed
→ Use Provisioned Throughput (EFS) or gp3 tuning
7. Common Mistakes (VERY IMPORTANT)
Avoid these in the exam:
❌ Using S3 for low-latency database
❌ Using sc1 for high-performance workloads
❌ Forgetting to configure IOPS in gp3
❌ Not enabling EBS optimization
❌ Ignoring throughput vs IOPS differences
8. Quick Revision Summary
- Latency → speed of response
- IOPS → number of operations
- Throughput → amount of data transfer
Storage Choices:
- S3 → scalable object storage
- EBS → high-performance block storage
- EFS → shared file storage
- FSx → specialized high-performance systems
Final Exam Strategy
When solving questions:
- Identify workload type (DB, file, analytics)
- Check performance requirement (IOPS, latency, throughput)
- Choose correct storage service
- Apply correct configuration (IOPS, throughput, mode)
