A Deep Dive into AWS Application Discovery
Service (ADS).
Scope:
- Architecture,
- Agent types,
- Data collection,
- Integrations,
- Migration workflows,
- Planning patterns,
- Limitations,
- Expert-level best practices.
Breakdown:
- Key Use Cases,
- Discovery Methods,
- ADS Architecture Overview,
- Data Collected by ADS,
- Integrations with Other AWS
Services,
- ADS in the Migration
Acceleration Program,
- Planning with Application
Dependency Mapping,
- Limitations,
- Best Practices (Expert Level),
- Data Export Capabilities,
- Recommended Architecture
Pattern,
- Final thoughts.
Intro:
- AWS Application Discovery Service (ADS) is a tool used during large-scale migration planning to collect detailed inventory, configuration, performance, and dependency data from on-premises environments to AWS.
- AWS Application Discovery Service (ADS) is a foundational component of the AWS Migration Acceleration Program (MAP) and is used to build migration plans, Total Cost of Ownership (TCO) models, wave planning, and application dependency maps.
ADS discovers key components:
- Servers (physical
& virtual)
- Processes
- Network dependencies
- Application components
- Utilization patterns
- Performance metrics (CPU, memory, I/O)
- Installed applications and versions
NB:
- This information feeds migration planning tools such as Migration Hub, AWS Migration Evaluator, and AWS Migration Hub Strategy Recommendations.
1. Key Use Cases
|
Use Case |
Description |
|
Application Dependency Mapping |
Identifies all
systems, services, ports, and flows involved in communication. |
|
Migration planning |
Build migration
waves, grouping interdependent servers. |
|
Right-sizing on AWS |
Uses actual
performance data to recommend optimal EC2 sizes. |
|
TCO estimation |
Combines
utilization with licensing/inventory data. |
|
Portfolio discovery |
Inventory of all
systems, versions, OS, and configurations. |
|
Modernization pathway identification |
Links with Strategy
Recommendations to analyze modernization candidates. |
NB:
ADS is
essential for enterprise migrations to avoid:
- Breaking dependencies during lift-and-shift
- Overprovisioning AWS resources
- Security issues due to unmanaged endpoints
- Incorrect grouping of servers
2. Discovery Methods
- ADS provides three ways to collect data:
A. Agent-Based Discovery
- Uses lightweight agents installed on Windows or Linux servers.
Collects:
- Hardware
specs
- Running
processes
- Open
network connections
- Port
mappings
- Inter-server
dependencies
- Performance
metrics (CPU, RAM, IOPS)
- Installed
applications
- Process-level
details (better than network-only)
Pros:
- Most
detailed dataset
- Dependency
mapping is highly accurate
- Required
for deep migration analysis
- Works on
physical & virtual machines
Cons:
- Requires
agent installation
- Harder
for restricted/locked-down servers
B. Agentless Discovery (using
VMware connector)
Runs as
an OVA inside vCenter.
Collects:
- VM
metadata (CPU, memory, disks)
- Host
relationships
- Network
info
- Performance
metrics (from hypervisor)
- No
OS-level process or dependencies
Pros:
- No agent
installation
- Enterprise-friendly
for VMware
- Quick
initial discovery
Cons:
- No
process-level or network mapping
- No
application identification
C. Data Import (CSV/XLSX)
Useful for:
- Legacy
environments
- Mainframes
- Unsupported
OS (Solaris, AIX, HP-UX)
- Manual
portfolio entries
Used for:
- TCO
estimates
- Migration
Hub portfolio completeness
- Wave
planning when ADS cannot collect data
3. ADS Architecture Overview
Main Components:
- On-premises agents / VMware Connector
- ADS Data Collectors
- Secure upload pipeline
- Migration Hub (central view)
- Migration Evaluator (TCO modeling)
- AWS Application Discovery APIs for integration
Data Flow:
- Agents/Connectors collect data locally
- Data is compressed and encrypted
- Sent to ADS backend in AWS
- Aggregated in ADS data store
- Visible in AWS Migration Hub
- Exportable to CSV/Jupyter/TCO tools
NB:
- ADS does not require inbound firewall rules — all
communication is outbound only
4. Data Collected by ADS
Server Inventory
- Hostname, OS, kernel
- CPU count, sockets
- Memory
- Storage (size & type)
- Network interfaces
Performance
Profiling
- CPU utilization (avg, peak, p95)
- Memory consumption
- Disk throughput (IOPS / MB/s)
- Network throughput
- Daily/weekly usage patterns
🌐 Network Dependency
Data
- Ports connecting IN/OUT
- Connection frequency
- Bandwidth
- Remote hosts
- Directional mapping
- Application traffic flows
Creates a full Application Dependency Graph (ADG).
Application Inventory
- Application name & version
- Installed packages
- Services running
- Middleware (Apache, IIS, JBoss, WebSphere)
- Database engines (SQL Server, Oracle, MySQL)
5. Integrations with Other AWS Services
|
Service |
Integration |
|
AWS Migration Hub |
Central repository
for all discovered assets. |
|
Migration Evaluator |
TCO modeling using
metrics. |
|
AWS MGN (Application Migration Service) |
Pre-migration
sizing. |
|
AWS SMS (Server Migration Service) |
Legacy integration. |
|
AWS Strategy Recommendations |
Modernization
analysis. |
|
Migration Hub Refactor Spaces |
Migration blueprint
generation. |
NB:
- ADS is most
powerful when coupled with Migration Hub, as it allows wave planning and dependency visualization.
6. ADS in the Migration Acceleration Program
(MAP)
ADS supports each MAP phase:
Phase 1: Assess
- Portfolio discovery
- Performance baseline
- Licensing inventory
- TCO estimation using real data
Phase 2: Mobilize
- Group workloads into migration waves
- Identify blockers
- Determine refactor vs rehost
Phase 3: Migrate and Modernize
- Provide sizing inputs to:
- AWS MGN
- SMS
- CloudEndure
- Validate dependency requirements during cutover
7. Planning with Application Dependency Mapping
- Dependency mapping is crucial to avoid migration breakage.
ADS identifies:
- Application tiers
- Web → App → DB
- Shared services (DNS, AD, NTP)
- Inter-app communication
- Legacy integration endpoints
- Batch jobs
This allows teams to:
- Build accurate migration wave groups
- Discover hidden dependencies
- Prevent outages
8. Limitations
Not everything is supported.
Discovery
Limitations:
- No support for AIX, Solaris, HP-UX (manual import required)
- VMware connector works only with vCenter (not standalone ESXi)
- Does not capture SMB shares, NFS mounts in detail
- No deep database schema discovery
- No layer-7 protocol intelligence
- Agents cannot be installed on locked-down OS
Functional
Limits:
- Data retention: typically 1 year
- No real-time analysis (data is near real-time but not streaming)
- Cannot detect business-level application structure automatically
9. Best Practices (Expert
Level)
A. Deploy ADS early
Start data collection at least 30–90 days before migration planning,
so:
- Full cyclic workloads are captured
- Seasonal peaks are visible
B. Use both agent + agentless
- Agentless for broad inventory
- Agent-based for deep analysis
C. Tag discovered applications
Using Migration Hub:
- map servers → applications
- ensure dependency accuracy
D. Gather 3–4 peak workload cycles
- Critical for accurate sizing.
E. Integrate with CMDB
- Export ADS data to enrich CMDB (ServiceNow/Atlassian/etc.).
F. Build migration waves based on:
- dependency groupings
- common maintenance windows
- business functions
- latency sensitivity
G. Secure ADS data
- Keep data private by locking down IAM roles.
10. Data Export Capabilities
ADS allows
exporting:
- Server inventory (CSV)
- Utilization metrics
- Dependency mapping data
- Application groupings
- Portfolio data
Exports are frequently used in:
- Excel/PowerBI dashboards
- Jupyter notebooks
- MAP TCO tools
- Hybrid architecture diagrams
11. Recommended Architecture Pattern
A modern migration
discovery architecture:
On-Prem Servers → ADS Agents → ADS Backend → Migration Hub → VMware Connector (optional) → TCO Engine → Migration Evaluator → Strategy Recommendations → Modernization Path
Final thoughts
- AWS Application Discovery Service (ADS) is a foundational tool in large-scale cloud migrations.
- AWS Application Discovery Service (ADS) ensures accurate planning, dependency analysis, capacity estimation,
and portfolio rationalization — preventing costly missteps during migration.
No comments:
Post a Comment