AWS Knowledge Hub

English

Tiếng Việt

English

Tiếng Việt

Appearance

Amazon EBS ​

Amazon Elastic Block Store (EBS) provides persistent block storage volumes for use with Amazon EC2 instances.

Overview ​

EBS volumes are network-attached storage that persist independently from the life of an instance. They provide high availability and durability, making them suitable for a wide variety of workloads.

Key Features ​

  • Persistent Storage: Data persists independently of instance life
  • High Performance: Low-latency, high-throughput block storage
  • Multiple Volume Types: Optimized for different use cases
  • Snapshots: Point-in-time backups stored in S3
  • Encryption: Built-in encryption at rest
  • Multi-Attach: Attach volumes to multiple instances (io1/io2)
  • Elastic Volumes: Modify volume size, type, and IOPS without downtime

Volume Types ​

General Purpose SSD (gp3) ​

  • Baseline Performance: 3,000 IOPS, 125 MB/s
  • Maximum Performance: 16,000 IOPS, 1,000 MB/s
  • Use Cases: Most workloads, boot volumes
  • Cost: Most cost-effective SSD option
  • Size: 1 GB to 16 TB

Provisioned IOPS SSD (io1/io2) ​

  • Maximum IOPS: 64,000 IOPS (io1), 256,000 IOPS (io2)
  • Maximum Throughput: 1,000 MB/s
  • Use Cases: I/O-intensive applications, databases
  • Cost: Higher cost, predictable performance
  • Size: 4 GB to 16 TB
  • Multi-Attach: Supported (io1/io2)

Throughput Optimized HDD (st1) ​

  • Maximum Throughput: 500 MB/s
  • Maximum IOPS: 500
  • Use Cases: Big data, data warehouses, log processing
  • Cost: Lower cost per GB
  • Size: 125 GB to 16 TB

Cold HDD (sc1) ​

  • Maximum Throughput: 250 MB/s
  • Maximum IOPS: 250
  • Use Cases: Less frequently accessed data
  • Cost: Lowest cost per GB
  • Size: 125 GB to 16 TB

Magnetic (standard) ​

  • Maximum IOPS: 40-200
  • Maximum Throughput: 40-90 MB/s
  • Use Cases: Legacy workloads
  • Cost: Low cost
  • Note: Legacy volume type, not recommended for new deployments

Core Concepts ​

Volumes ​

Block-level storage devices that can be attached to EC2 instances. Volumes are Availability Zone-specific.

Snapshots ​

Point-in-time copies of EBS volumes stored in S3. Snapshots are incremental and region-specific.

Volume Attachments ​

The association between an EBS volume and an EC2 instance. Volumes must be in the same AZ as the instance.

Elastic Volumes ​

Ability to modify volume configuration (size, type, IOPS) without detaching or creating a new volume.

Performance Characteristics ​

IOPS (Input/Output Operations Per Second) ​

  • Measures read/write operations per second
  • Important for database and transactional workloads
  • Can be provisioned independently (io1/io2)

Throughput ​

  • Measures data transfer rate (MB/s)
  • Important for large file transfers and sequential workloads
  • Depends on volume type and size

Latency ​

  • Time for a single I/O operation
  • Lower latency = better performance
  • SSD volumes have lower latency than HDD

Snapshots ​

Automated Snapshots ​

  • Create snapshots on a schedule using Data Lifecycle Manager
  • Automate backup and retention policies
  • Cost-effective backup strategy

Manual Snapshots ​

  • User-initiated point-in-time backups
  • Retained until explicitly deleted
  • Can be shared across accounts and regions

Snapshot Features ​

  • Incremental: Only changed blocks are stored
  • Encrypted: Snapshots of encrypted volumes are encrypted
  • Copyable: Can copy snapshots across regions
  • Shareable: Can share snapshots with other accounts

Encryption ​

Encryption at Rest ​

  • AWS KMS Integration: Uses AWS KMS for key management
  • Automatic: Can enable encryption by default
  • Performance: No performance impact
  • Snapshots: Encrypted volumes create encrypted snapshots

Encryption in Transit ​

  • Data encrypted when moving between instance and volume
  • No additional configuration required
  • Uses AWS encryption protocols

Multi-Attach ​

Supported Volume Types ​

  • io1 and io2 volumes only
  • Up to 16 instances per volume
  • All instances must be in the same AZ

Use Cases ​

  • Clustered applications
  • High availability applications
  • Concurrent write workloads

Considerations ​

  • Requires cluster-aware file systems
  • Applications must handle concurrent access
  • Not suitable for all workloads

Best Practices ​

  • Right-Size Volumes: Match volume type to workload requirements
  • Use gp3 for Most Workloads: Best balance of cost and performance
  • Enable Encryption: Encrypt sensitive data at rest
  • Regular Snapshots: Implement automated backup strategy
  • Monitor Performance: Use CloudWatch to monitor volume metrics
  • Use Elastic Volumes: Modify volumes without downtime
  • Delete Unused Snapshots: Reduce storage costs
  • Use Provisioned IOPS for Databases: For consistent performance
  • Implement Lifecycle Policies: Automate snapshot management

Use Cases ​

  • Boot Volumes: Root volumes for EC2 instances
  • Databases: High-performance database storage
  • Applications: Application data and logs
  • Big Data: Data processing and analytics
  • Backup Storage: Snapshot-based backup solutions

Cost Optimization ​

  • Use gp3 instead of gp2 for better price/performance
  • Right-size volumes to match actual usage
  • Delete unused snapshots
  • Use st1/sc1 for appropriate workloads
  • Implement snapshot lifecycle policies
  • Monitor and optimize unused volumes
  • Use Reserved Capacity for predictable workloads

Limits ​

  • Volume Size: 1 GB to 16 TB (varies by type)
  • Volumes per Instance: Up to 40 volumes
  • Total Storage per Instance: Up to 300 TB
  • Snapshots: Unlimited (charged for storage)
  • IOPS: Up to 256,000 (io2)