Building blocks
Design goal
Produce rightsized, defensible Azure migration recommendations that reflect real workload behavior and application dependencies rather than one-time server specifications.
Success criteria
- Create an inventory
- Collect configuration and performance evidence
- Identify readiness blockers
- Generate rightsized targets and migration waves
Architecture
Use appliance-based or import-based discovery depending on source access and depth required; group workloads by application before creating assessments.
Architecture flow
- Discover or import inventory
- Collect configuration/performance data
- Group related workloads
- Run assessment
- Review readiness, sizing, cost, and dependencies
- Create migration waves
Architecture decisions
Prefer deeper discovery when available
WhyAppliance-based discovery provides recurring configuration/performance data and a richer basis for assessment than a one-time inventory import.
Trade-offContinuous discovery improves sizing confidence and dependency context, but requires appliance deployment, credentials, connectivity, and a longer observation window.
Assess applications as groups
WhyServer-by-server assessment can miss dependencies that determine migration sequencing and outage planning.
Trade-offApplication grouping produces better migration sequencing, but requires ownership and dependency information that server inventories often do not contain.
Security
Protect the control and data paths deliberately. Use least-privilege discovery credentials, control appliance network access, and treat collected inventory as sensitive infrastructure data.
Cost drivers
- Assessment target sizes
- Licensing assumptions
- Disk performance assumptions
- Operating schedule
- Network egress/ingress design after migration
Design assumptions
- The source environment can provide reliable inventory data
- A representative performance window is available for performance-based sizing
Implementation plan
- Define the assessment scope, ownership, target region, licensing assumptions, and whether discovery uses an appliance or imported inventory.
- Deploy the appliance or prepare import data, then reconcile discovered servers with the source platform before sizing decisions are made.
- Collect enough performance history to capture representative peaks and seasonality instead of relying on a short observation window.
- Group workloads into applications and review dependencies before using assessment results to drive migration-wave sequencing.
- Run alternative sizing and target scenarios, then review readiness, outliers, licensing, and cost assumptions with workload owners.
Validate the design
- Reconcile discovered inventory with the source platform and explain missing, duplicate, or intentionally excluded systems.
- Review performance outliers and low-confidence data before accepting rightsizing recommendations.
- Confirm application groupings and critical dependencies with owners before turning assessment output into migration waves.
- Compare at least one alternative sizing/licensing scenario for major cost outliers and document the chosen assumption.