EBS Volumes

Amazon EBS (Elastic Block Store)

Key Characteristics

• Annual failure rate: < 0.2% (vs. 4% for commodity drives)

• Availability target: 99.999%

• Single-AZ replication

• Snapshot support with S3 storage

• Cross-region snapshot copying available

RAID Configuration Options

RAID 0 (Striping)

In a RAID 0 configuration, the data is sharded (striped) across multiple volumes. This means that consecutive blocks of data are written alternately to different drives/volumes in the array.

• No redundancy

• Data distributed across multiple drives

• Advantages:

  • Improved read/write performance

  • Full capacity utilization

  • Double throughput

• Use Case:

  • When performance is priority over redundancy

RAID 1 (Mirroring)

In RAID 1 (mirroring), the data is not sharded but rather fully duplicated across volumes. Each write operation is performed on all volumes in the array, creating exact copies.

• Full redundancy with duplicate copies

• Characteristics:

  • 50% capacity reduction

  • Slightly slower than RAID 0

  • Same throughput as single volume

• Use Case:

  • When data redundancy is critical

RAID 5 (Not Recommended for AWS)

• Minimum 3 drives required

• Uses parity for redundancy

• Not recommended due to:

  • Network I/O overhead

  • 20-30% IOPS consumption for parity

  • Performance impact

RAID 6 (Not Recommended for AWS)

• Minimum 4 drives required

• Dual parity methods

• Can survive two drive failures

• Not recommended for same reasons as RAID 5

EBS Performance Examples

1. Single Volume Configuration

   • 1 TB volume

   • 4000 IOPS

   • 500 Mbps throughput

   • 1 TB usable space

2. RAID 0 Configuration

   • 2 x 500 GB volumes

   • 8000 total IOPS

   • 1000 Mbps throughput

   • 1 TB usable space

3. RAID 1 Configuration

   • 2 x 500 GB volumes

   • 4000 total IOPS

   • 500 Mbps throughput

   • 500 GB usable space

Amazon S3 Storage Classes

Standard Storage

• Availability: 99.99%

• Multi-AZ durability

• Durability: 11 nines (99.999999999%)

• Immediate access

• Use Case: Frequently accessed data

Standard Infrequent Access (IA)

• Slightly lower availability

• Multi-AZ durability

• 8.7 hours/year potential unavailability

• Lower cost than Standard

• Use Case: Less frequently accessed data

One-Zone Infrequent Access

• Single-AZ durability

• 1.8 days/year potential unavailability

• Maintains 11 nines durability

• Lowest cost for IA tier

• Use Case: Reproducible, infrequently accessed data

Amazon EFS (Elastic File System)

Features

• NFS implementation

• Multi-AZ redundancy

• Scalable file sharing

• Automatic replication

Architecture Benefits

• Multiple mount points across AZs

• Continued operation during AZ failure

• Automatic scaling support

• Integrated with EC2 auto-scaling

Additional Storage Solutions

Amazon Storage Gateway

• Ideal for on-premises to AWS migration

• Supports ongoing synchronization

• Provides hybrid storage capabilities

• Good for offsite backups

AWS Snowball

• Suited for batch data transfers

• Physical data transport solution

• Various capacity options

• Secure data migration

Best Practices

EBS Implementation

• Use snapshots for backup

• Consider RAID 0 or 1 only

• Monitor IOPS utilization

• Plan for AZ failures

S3 Usage

• Choose appropriate storage class

• Utilize lifecycle policies

• Consider cross-region replication

• Monitor availability metrics

EFS Deployment

• Implement multi-AZ mount points

• Plan for redundancy

• Monitor throughput

• Consider backup strategies

Availability Considerations

Single-AZ Services

• Plan for AZ failures

• Implement snapshot strategies

• Consider replication options

• Monitor service health

Multi-AZ Services

• Leverage built-in redundancy

• Implement cross-AZ architecture

• Plan for regional failures

• Consider cross-region options