RTO、RPO 与 DR 层级

理解 RTO 和 RPO

# RTO and RPO definitions:
# RTO = max time system can be DOWN
#   Example: RTO=4h means restore within 4 hours
#
# RPO = max data LOSS acceptable
#   Example: RPO=1h means no more than 1 hour of data lost
#
# Lower RTO and RPO = more expensive DR strategy
# Higher RTO and RPO = cheaper but more business impact

四种 DR 层级

# DR Strategy comparison:
# Strategy          | RTO      | RPO      | Cost
# Backup & Restore  | Hours    | Hours    | Lowest
# Pilot Light       | Minutes+ | Minutes  | Low
# Warm Standby      | Minutes  | Seconds  | Medium
# Active-Active     | ~0       | ~0       | Highest

Backup and Restore 策略

# Create AWS Backup plan for RDS
aws backup create-backup-plan \
  --backup-plan '{
    "BackupPlanName": "daily-backup",
    "Rules": [{
      "RuleName": "daily",
      "TargetBackupVaultName": "dr-vault",
      "ScheduleExpression": "cron(0 5 ? * * *)",
      "StartWindowMinutes": 60,
      "CompletionWindowMinutes": 180,
      "Lifecycle": {
        "DeleteAfterDays": 35
      },
      "CopyActions": [{
        "DestinationBackupVaultArn": "arn:aws:backup:us-west-2:123:backup-vault:dr-vault"
      }]
    }]
  }'

Pilot Light 策略

# Pilot Light: what runs in DR region at all times
# - RDS Read Replica (continuously replicated)
# - Core VPC/networking infrastructure
# - Route 53 DNS (inactive until failover)

# What is NOT running (launched during failover):
# - EC2 application servers
# - ELB (or dormant)

# Failover steps:
# 1. Promote RDS Read Replica to standalone
# 2. Scale up EC2 instances from launch template
# 3. Update Route 53 to point to DR region

Warm Standby 策略

# Warm Standby: DR region runs scaled-down version
# Production:  10 EC2 instances (ASG min=10, max=50)
# DR Standby:  2 EC2 instances  (ASG min=2,  max=50)

# During failover:
# 1. Route 53 health check fails for primary
# 2. DNS switches to DR ALB
# 3. ASG in DR scales up from 2 to 10+
# 4. Promote Aurora Global DB secondary
# Total failover time: ~5-15 minutes

Multi-Site Active-Active 策略

# Active-Active: full capacity in both regions
# us-east-1:  ASG 10 instances (serving ~50% traffic)
# eu-west-1:  ASG 10 instances (serving ~50% traffic)

# Route 53 weighted routing:
# us-east-1: weight=50
# eu-west-1: weight=50
# Both records have health checks

# On us-east-1 failure:
# Health check fails -> Route 53 removes us-east-1
# eu-west-1 receives 100% traffic
# ASG in eu-west-1 scales up automatically

RPO 与数据复制技术

# RPO requirements mapped to replication technology:
# RPO = 0:        RDS Multi-AZ (synchronous)
# RPO < 1 second: Aurora Global Database
# RPO < 1 minute: DynamoDB Global Tables
# RPO < 15 min:   RDS Read Replica
# RPO < 1 hour:   AWS Backup hourly schedule
# RPO < 24 hours: AWS Backup daily schedule

无服务器架构的 DR

# Deploy Lambda to multiple regions with CDK
# cdk.json environment configuration:
{
  'primary': {
    'account': '123456789',
    'region': 'us-east-1'
  },
  'dr': {
    'account': '123456789',
    'region': 'us-west-2'
  }
}

# Deploy to both:
# cdk deploy --context env=primary
# cdk deploy --context env=dr

RTO 与成本权衡示例

# DR cost justification formula:
# Annual cost of DR infrastructure
# vs
# Expected annual outage cost
#   = P(outage) x downtime_duration x cost_per_minute
#
# Example:
# P(annual outage) = 0.1 (10% chance per year)
# downtime = 8 hours = 480 minutes
# cost = $1000/min
# Expected loss = 0.1 x 480 x $1000 = $48,000/year
#
# If warm standby costs $30,000/year -> worth it

DR 测试与文档

# DR drill checklist:
# 1. Notify stakeholders (planned drill)
# 2. Initiate failover (Route 53 health check override)
# 3. Measure time from trigger to traffic in DR region (RTO)
# 4. Measure data consistency between regions (RPO)
# 5. Test all critical application functions in DR
# 6. Failback to primary region
# 7. Document actual RTO/RPO vs target
# 8. Update runbooks with lessons learned

合规性与 DR 要求

# AWS Config rule to check RDS backup retention
aws configservice put-config-rule \
  --config-rule '{
    "ConfigRuleName": "rds-backup-enabled",
    "Source": {
      "Owner": "AWS",
      "SourceIdentifier": "DB_INSTANCE_BACKUP_ENABLED"
    },
    "InputParameters": "{\"backupRetentionMinimum\":\"7\"}"
  }'

快速检查

课程回顾