cloudnative wiki

Application Discovery Service

5 min read

Application Discovery Service

Application Discovery Service (ADS) helps you discover and understand your on-premises environment before migration. It collects infrastructure data (servers, utilization, network dependencies) that feeds into Migration Hub for planning and tracking.

Discovery Methods

Agentless Discovery (VMware)

No software installed on source servers — a collector VM interrogates vCenter:

# Start an agentless collector scan
aws discovery start-agentless-collector-scan \
  --region us-east-1 \
  --collectors ['arn:aws:discovery:us-east-1:123456789012:agentless-collector/collector-abcdef0']
 
# Describe collectors
aws discovery list-agents --region us-east-1
 
# Get discovered servers
aws discovery list-servers --region us-east-1
 
# Get server details
aws discovery describe-agents --agent-ids arn:aws:discovery:us-east-1:123456789012:agent/agent-abcdef0

What it collects:

  • VM name, guest OS, version
  • vCPU, memory, disk sizes
  • Network configuration (IP addresses, MAC addresses)
  • Resource utilization (CPU, memory, disk over time)
  • VMware cluster and resource pool membership

Agent-Based Discovery

Install the discovery agent on each server for deeper visibility:

# Download and install the agent (Linux)
wget https://s3.amazonaws.com/aws-discovery-agent/linux/latest/aws-discovery-agent.tar.gz
tar -xzf aws-discovery-agent.tar.gz
sudo ./install
 
# Configure the agent
sudo /opt/aws/discovery agent configure --region us-east-1
 
# Check agent status
sudo /opt/aws/discovery/bin/AWSDiscoveryAgent status
 
# Check logs
sudo tail -f /var/log/aws/discovery/discovery-agent.log

What it collects (beyond agentless):

  • Running processes and services
  • Application names and versions
  • Network connections between servers (dependency mapping)
  • User accounts on the server
  • More detailed performance metrics

Agent Deployment

VMware Agentless Collector

# 1. Download the collector OVA from ADS console
# 2. Deploy in VMware (2 vCPU, 8GB RAM, 100GB disk)
# 3. Configure with your AWS credentials (IAM role or access key)
# 4. Configure vCenter credentials (read-only is sufficient)
# 5. Let it run — it will continuously collect for 2-4 weeks

Requirements:

  • Outbound HTTPS to AWS (port 443)
  • Access to vCenter API (read-only)
  • NTP configured (accurate timestamps are important)

Linux Agent

# Install on RHEL/CentOS/Amazon Linux
sudo yum install ./aws-discovery-agent.rpm
 
# Install on Ubuntu/Debian
sudo dpkg -i ./aws-discovery-agent.deb
 
# Configure after install
sudo /opt/aws/discovery/bin/configure.py \
  --region us-east-1 \
  --agent-id my-agent-group-1

Windows Agent

  1. Download the installer from ADS console
  2. Run the installer (MSI package)
  3. Open Services → AWSDiscoveryAgent → set to Automatic
  4. Edit config at C:\Program Files\AWS\AWSDiscoveryAgent\config.ini with your region

Data Collected

Server Information

{
  "serverId": "server-abc123",
  "hostname": "web-prod-01",
  "osVersion": "RHEL 8.5",
  "osArchitecture": "x86_64",
  "cpuCores": 8,
  "memorySizeMB": 16384,
  "diskSizeMB": 512000,
  "ipAddresses": ["10.0.1.50", "192.168.1.100"],
  "macAddresses": ["00:0C:29:AB:CD:EF"],
  "vmwareConfig": {
    "clusterName": "production",
    "resourcePool": "web-tier",
    "vCenterId": "vc-12345"
  }
}

Network Dependencies

{
  "sourceServerId": "server-abc123",
  "connections": [
    {
      "destinationIp": "10.0.1.100",
      "destinationPort": 3306,
      "protocol": "TCP",
      "processName": "mysqld"
    },
    {
      "destinationIp": "10.0.1.101",
      "destinationPort": 6379,
      "protocol": "TCP",
      "processName": "redis-server"
    }
  ]
}

This dependency map is critical for understanding what must be migrated together and what the blast radius is when cutting over.

Utilization Data

{
  "serverId": "server-abc123",
  "utilizationSamples": [
    {
      "timestamp": "2024-06-15T00:00:00Z",
      "cpuPercent": 35.2,
      "memoryPercent": 72.1,
      "diskReadBytesPerSec": 125000,
      "diskWriteBytesPerSec": 85000,
      "networkInBytesPerSec": 2500000,
      "networkOutBytesPerSec": 3500000
    }
  ]
}

This data feeds into Migration Evaluator for right-sizing recommendations.

Exporting Data

# Export all discovered data as CSV
aws discovery get-discovered-resource-summary --region us-east-1
 
# Export to S3
aws discovery export-descriptions \
  --output-format CSV \
  --s3-destination '{
    "Bucket": "my-discovery-data",
    "Prefix": "export/2024-06/"
  }'
 
# List exports
aws discovery list-export-configurations --region us-east-1

Integration with Migration Hub

Discovered servers can be imported into Migration Hub as applications:

# Register an application
aws migrationhub create-application \
  --name "order-service" \
  --description "Order processing service"
 
# Associate discovered servers with application
aws discovery associate-discovered-resource \
  --identifier '{"source": "AWSApplicationDiscoveryService", "id": "server-abc123"}' \
  --destination-region us-east-1

This gives you a migration plan with all servers tracked in Migration Hub, ready for MGN or DMS migration.

Querying Discovery Data

# List all servers
aws discovery list-servers
 
# Filter by OS
aws discovery query \
  --expression "SELECT * FROM AWSDISCOVERY WHERE osType = 'RHEL'"
 
# Get servers with high CPU utilization
aws discovery query \
  --expression "SELECT serverId, hostname, cpuCores FROM AWSDISCOVERY WHERE cpuCores > 4"
 
# Get network dependencies for a specific server
aws discovery list-server-neighbors \
  --server-id server-abc123 \
  --region us-east-1

Important: The query API uses SQL-like syntax against your discovery data stored in AWS (not a real SQL database — just a query interface).

Monitoring

# Check agent status
aws discovery describe-agents --agent-ids agent-abc123
 
# CloudWatch metrics for ADS
# - HostAgentHeartbeat (is the agent communicating)
# - DataSize (how much discovery data is being collected)

Limitations

  • Agentless: VMware only (no Hyper-V, physical, other hypervisors)
  • Agent: Supports Windows Server and Linux (RHEL, CentOS, Ubuntu, SUSE, Debian)
  • Data retention: Discovery data retained for 90 days (can export before deletion)
  • No migration execution: ADS only discovers, doesn’t migrate. Use MGN, DMS, DataSync for actual migration.
  • AWS regions: Discovery data can only be exported to the same region where it was collected

Use in Migration Planning

Phase 1: Discovery (ADS)
  → Collect server inventory, utilization, dependencies

Phase 2: Assessment (Migration Evaluator)
  → Analyze discovery data, generate TCO report, right-sizing recommendations

Phase 3: Planning (Migration Hub)
  → Group servers into applications, assign migration strategy

Phase 4: Execution (MGN, DMS, DataSync)
  → Execute the migration per plan

Phase 5: Validation
  → Verify applications work post-migration

ADS is the starting point. Without knowing what you have, you can’t plan the migration or build an accurate business case.

Graph View