Azure Migrate: Discovery and Assessment

Azure Migrate is a centralised hub in the Azure portal that provides tools for discovering, assessing, and migrating on-premises servers, databases, web apps, and virtual desktops to Azure. It integrates with both Microsoft-built tools and a growing ecosystem of independent software vendor (ISV) tools such as Cloudamize and Movere. Azure Migrate stores all discovery and assessment data in a dedicated project resource within a resource group.

Discovery begins by deploying the Azure Migrate appliance — a lightweight VM that you run in your on-premises environment. The appliance continuously collects metadata, performance counters, and dependency data from VMware vCenter, Microsoft Hyper-V, or physical servers and streams them securely to the Azure Migrate service over HTTPS. A single appliance can discover up to 10,000 VMware VMs or 5,000 Hyper-V VMs.

The appliance collects rich metadata including: server inventory (OS, version, installed software, CPU, RAM, disk count), performance data (CPU and memory utilisation over up to 12 months), network dependencies (which server talks to which), and storage configuration (disk sizes and IOPS). For databases, a separate SQL Server discovery credential allows the appliance to also enumerate SQL instances and databases.

After discovery, you create an Azure VM assessment in the Azure Migrate portal. You configure the assessment with: target region, VM sizing criterion (performance-based or as on-premises), reserved instance discount option, and comfort factor (a buffer multiplier, default 1.3). The assessment analyses performance data and recommends the right Azure VM SKU and estimated monthly cost for each server.

Each assessed VM receives an Azure readiness rating: Ready, Ready with conditions, Not ready, or Readiness unknown. Common reasons for Not ready include: boot disks larger than Azure supports, unsupported operating system versions, or more than 64 data disks attached. The assessment report details every readiness issue and provides actionable remediation steps.

The performance-based sizing criterion examines the peak utilisation of CPU and memory over the collection period and multiplies it by the comfort factor to select the smallest Azure VM SKU that can handle the workload. For example, a server consistently using 20% of 16 vCPUs (3.2 vCPUs effective) with a 1.3 comfort factor requires a VM SKU offering at least 4.16 vCPUs — so the assessment recommends the next-tier SKU with 4 or 8 vCPUs.

Dependency analysis shows which servers communicate with each other, helping you group interdependent apps into migration waves. Azure Migrate supports two dependency modes: agentless (uses vCenter data, no software to install, less granular) and agent-based (installs the Log Analytics Dependency agent and MMA/AMA on each VM for process-level connection details). Dependency maps are visualised directly in the Azure portal.

Azure Migrate also assesses ASP.NET web applications for migration to Azure App Service and SQL Server databases for migration to Azure SQL Database, Azure SQL Managed Instance, or SQL Server on Azure VM. The SQL assessment identifies deprecated features, unsupported collations, and compatibility issues, giving each database a readiness status and recommending the optimal Azure SQL target based on features used.

The assessment generates a detailed monthly cost estimate for each workload in the target Azure region, broken down by compute, storage, and network egress. You can toggle between pay-as-you-go and 1-year/3-year reserved instance pricing to model cost savings. Exporting the report to Excel allows you to build a total business case comparing on-premises TCO with projected Azure spend over a 3-year horizon.

Azure Migrate assigns a confidence rating (1–5 stars) to each assessment based on the percentage of performance data points collected. An assessment with less than 20% data available gets 1 star and may significantly over- or under-size VMs. Best practice is to collect at least 30 days of performance data before generating an assessment to achieve the maximum 5-star confidence rating and reliable cost estimates.

Assessments are point-in-time snapshots — they do not update automatically. As you gather more performance data or change your assumptions (target region, discount model, comfort factor), you should create a new assessment to reflect the updated inputs. Azure Migrate lets you compare multiple assessments side by side so you can model different scenarios (for example, East US vs. North Europe, or pay-as-you-go vs. reserved) before committing to a migration plan.

Test your understanding of Microsoft Azure Fundamentals (AZ-900) concepts from this lesson.

In this lesson you learned: the Azure Migrate appliance discovers on-premises servers and streams performance data, server assessments recommend Azure VM SKUs and estimate monthly costs, and dependency analysis groups interdependent servers into migration waves. Next up we execute a Rehost migration using Azure Migrate's replication engine.