Application Migration Service (MGN)

MGN is AWS’s modern lift-and-shift (rehost) service. It continuously replicates your on-premises servers to AWS, allowing you to cut over with minimal downtime. It replaces the older Server Migration Service (SMS).

How It Works

Source Server                    MGN Service                   Target AWS
(Windows/Linux)                  (managed)                    (EC2)
    │                                │                            │
    ├─ Install MGN Agent ──────────→ │                            │
    │                                │                            │
    ├─ Continuous Replication ──────→ │ (S3 staging + EBS snapshots)
    │                                │                            │
    │                                │─── Test Launch ────────────→│ (validation)
    │                                │                            │
    │                                │─── Final Cutover ─────────→│ (production)

Replication process:

Install MGN agent on source server
Agent continuously replicates block-level changes to a staging area in your AWS account (S3 bucket + EBS snapshots)
When ready, initiate test launch or final cutover
MGN converts the replicated disk to an EC2 instance and boots it

Key Concepts

Source Server

A source server is a server (physical or virtual) that you’ve registered with MGN. Each source server has:

Replication configuration: VPC, subnet, security group for the replicated instance
Agent status: Online/Offline, last sync time
Lifecycle: PENDING, TESTING, MIGRATING, CUTOVER, CUTOVER_COMPLETE

Wave

A wave is a group of source servers that you migrate together as a unit. Waves let you coordinate cutover for multi-server applications.

# Create a wave
aws mgn create-wave \
  --account-id 123456789012 \
  --name "web-tier-wave-1" \
  --description "Web tier servers, cutover planned for Friday night"
 
# Add servers to wave
aws mgn add-source-servers-to-wave \
  --wave-id w-1234567890abcdef0 \
  --source-server-ids s-1234567890abcdef0 s-0987654321fedcba0
 
# Track wave status
aws mgn describe-waves --wave-id w-1234567890abcdef0

Cutover Modes

Test launch: Non-destructive launch of the replicated server. You can boot it, validate it works, then terminate. Original server keeps replicating.
Final cutover: Actual production cutover. After cutover, the original server is shut down and replication stops.

Installing the Agent

Linux

# Download the agent installer
# Get the installer URL from: MGN Console → Source Servers → Install Agent
wget https://aws-mgn-agent-bucket.s3.amazonaws.com/latest/linux/aws-replication-installer.ini
 
# Edit the installer config
# Add your AWS region and source server ID from the MGN console
cat > aws-replication-installer.ini << 'EOF'
[DEFAULT]
aws_region = us-east-1
source_server_id = s-1234567890abcdef0
bandwidth_throttle = 0
use_public_ip = false
EOF
 
# Run the installer
sudo ./aws-replication-installer.py

Windows

Download the agent installer (.msi) from MGN console
Run the installer on the Windows server
Enter the Source Server ID and AWS region when prompted
The agent runs as a Windows service (AWSReplicationService)

Agent Configuration

# Check agent status on Linux
sudo systemctl status aws-replication-agent
 
# Check agent logs
sudo tail -f /var/log/aws-replication-agent.log
 
# Pause/resume replication (for maintenance windows)
sudo /aws-replication-agen pause
sudo /aws-replication-agen resume
 
# Uninstall agent (after cutover complete)
sudo /aws-replication-agen uninstall

Replication Settings

# Configure replication (e.g., throttle bandwidth to 50 Mbps)
aws mgn update-source-server-replication-configuration \
  --source-server-id s-1234567890abcdef0 \
  --replication-configuration '{
    "bandwidthThrottling": 50,
    "dataReplicationRootHash": null,
    "enableReBoot": true,
    "replicatedServerDisks": [{
      "diskTag": "C:",
      "diskPadding": 0
    }],
    "replicationServerSecurityGroupIds": ["sg-0123456789abcdef0"],
    "replicationServerSubnetId": "subnet-0123456789abcdef0",
    "replicationServerSecurityGroupIds": ["sg-0123456789abcdef0"],
    "serverId": "s-1234567890abcdef0",
    "targetInstanceTypeRightSizingMethod": "AUTO"
  }'

Key settings:

bandwidthThrottling: Limit replication bandwidth (0 = unlimited)
enableReBoot: MGN can reboot the source if needed for replication (Windows only)
targetInstanceTypeRightSizingMethod: AUTO = right-size based on utilization, MANUAL = specify type

Cutover Workflow

# 1. Check replication status (should be "Live" before cutover)
aws mgn describe-source-servers \
  --filters "Name=source-server-id,Values=s-1234567890abcdef0"
# Look for: lastSyncedTime, estimatedSyncDuration, volumeStatus
 
# 2. Initiate test launch
aws mgn start-test-launch \
  --source-server-id s-1234567890abcdef0 \
  --target-instance '{
    "instanceType": "t3.medium",
    "vpcSecurityGroupIds": ["sg-0123456789abcdef0"],
    "subnetId": "subnet-0123456789abcdef0"
  }'
 
# 3. Wait for instance to boot (check status)
aws mgn describe-source-servers --source-server-id s-1234567890abcdef0
# Lifecycle should show: TESTING
 
# 4. Validate the test instance (application works, connectivity OK)
# 5. Terminate test instance when validation done
aws mgn terminate_target_instance \
  --source-server-id s-1234567890abcdef0
 
# 6. When ready for final cutover
aws mgn start-cutover \
  --source-server-id s-1234567890abcdef0 \
  --target-instance '{
    "instanceType": "t3.medium",
    "vpcSecurityGroupIds": ["sg-0123456789abcdef0"],
    "subnetId": "subnet-0123456789abcdef0"
  }'
 
# 7. Verify cutover complete
aws mgn describe-source-servers --source-server-id s-1234567890abcdef0
# Lifecycle should show: CUTOVER_COMPLETE

Wave Cutover

For multi-server applications, cut over the entire wave at once:

# Start wave cutover
aws mgn start-wave-cutover \
  --wave-id w-1234567890abcdef0 \
  --target-instance '{
    "targetInstanceTypeRightSizingMethod": "AUTO"
  }'
 
# Monitor wave cutover progress
aws mgn describe-wave-cutovers --wave-id w-1234567890abcdef0
 
# Check individual server status
aws mgn describe-source-servers --filters "Name=wave-id,Values=w-1234567890abcdef0"

Post-Launch Validation

MGN can run validation scripts after launch to verify the server is working:

# Create a post-launch validation template
aws mgn create-launch-configuration-template \
  --post-launch-configuration '{
    "targetInstance": {
      "instanceType": "t3.medium",
      "vpcSecurityGroupIds": ["sg-0123456789abcdef0"],
      "subnetId": "subnet-0123456789abcdef0"
    },
    "validation": {
      "commands": ["powershell -File C:\\validate.ps1"]
    }
  }'
 
# Or run validation via Lambda after cutover
# MGN can trigger a Lambda on cutover completion for custom validation

Monitoring

# CloudWatch metrics for MGN
# - Progress (percentage of replication complete)
# - DataTransferred (bytes replicated)
# - Lag (time behind source in seconds)
# - DiskUtilization (source disk usage)
# - NetworkBandwidthThrottle (actual bandwidth used)
 
# Check replication lag
aws mgn describe-source-server-replication-configuration \
  --source-server-id s-1234567890abcdef0
# Look at: lastSyncedTime, lagDuration

If lag is growing (not shrinking):

Network bandwidth is too low for the rate of data change
Increase bandwidthThrottling limit
Consider compressing replication traffic (enable compression in config)

Network Requirements

For MGN agent to work:

Outbound HTTPS (443) to MGN service endpoints
Outbound HTTPS to S3 endpoints (for staging data)
Inbound from replication server (for initial data transfer)

Ports required:

TCP 443 (HTTPS) — to AWS services
TCP 1500 — replication data transfer from source to replication server
UDP 1500 — heartbeat (optional, for faster failover)

Staging Area

MGN uses an S3 bucket and EBS snapshots as a staging area:

Replication data goes to S3 (encrypted with KMS)
EBS snapshots are created from S3 data
Staging area is cleaned up automatically after cutover

You can customize the staging area location (dedicated S3 bucket per account for cost tracking).

Cost

Per-source server per hour: ~$0.026/hr (varies by region)
S3 and EBS costs: For the staging area (temporary, deleted after cutover)
EBS snapshot costs: Until snapshots are deleted post-cutover

Cost tip: After cutover is complete, immediately clean up:

# Delete source server from MGN (stops billing)
aws mgn delete-source-server --source-server-id s-1234567890abcdef0
 
# Delete staging S3 bucket contents (stops S3 charges)
aws s3 rm s3://mgn-staging-bucket --recursive

Limitations

Supported sources: Physical servers, VMware ESXi, Hyper-V, AWS (EC2-to-EC2 migration), Azure VMs, Google Cloud VMs
Windows versions: 2008 R2, 2012, 2012 R2, 2016, 2019, 2022
Linux versions: RHEL, CentOS, Ubuntu, Debian, SUSE, Oracle Linux
Maximum source disks: 20 per server
Maximum source disk size: 16TB per disk
Replication to different VPC/Region: Not directly supported — replicate to same region first, then use VM Import/Export or create AMI and copy

Comparison: MGN vs SMS

Feature	MGN	SMS (deprecated)
Continuous replication	Yes	Yes
Wave management	Yes	No
Test launch	Yes	No
Agent-based	Yes	Agent-based
Status	Active development	No new features, eventually retired
Cutover modes	Test + Final	Direct cutover only

cloudnative wiki

Explorer

Application Migration Service (MGN)

Application Migration Service (MGN)

How It Works

Key Concepts

Source Server

Wave

Cutover Modes

Installing the Agent

Linux

Windows

Agent Configuration

Replication Settings

Cutover Workflow

Wave Cutover

Post-Launch Validation

Monitoring

Network Requirements

Staging Area

Cost

Limitations

Comparison: MGN vs SMS

Graph View

Table of Contents