Why the AWS-specific detail matters
A migration strategy is largely provider-neutral: discover, classify, plan waves, build a landing zone, migrate, validate, decommission. What differs by provider is the tooling available at each stage, and using the native tooling well is frequently the difference between a smooth migration and a manual, error-prone one.
Discovery: AWS Application Discovery Service
AWS Application Discovery Service collects configuration, performance and dependency data from on-premise servers, either agentlessly (via a virtual appliance polling hypervisor metrics) or with an agent installed per server for deeper process-level dependency mapping. The output — real CPU, memory and network utilisation, plus a dependency map — is the foundation the rest of the migration plan is built on, and it is materially better than relying on documentation or memory.
Landing zone: AWS Control Tower and Organizations
AWS Control Tower automates the setup of a multi-account landing zone: a management account, dedicated log archive and audit accounts, and guardrails (service control policies) that enforce baseline security and compliance rules across every account created afterward. Building this by hand is possible but slow and error-prone; Control Tower encodes AWS best practice for account structure so new accounts inherit sane defaults automatically.
- AWS Organizations provides the underlying multi-account structure and consolidated billing.
- IAM Identity Center (formerly AWS SSO) centralises human access across every account rather than managing IAM users per account.
- Service control policies set hard boundaries — for example, blocking regions that are out of scope for a compliance requirement — that individual account administrators cannot override.
Moving the workloads: DMS, MGN and Snow family
The right migration tool depends on what is moving:
- AWS Database Migration Service (DMS) replicates databases with minimal downtime, supporting both homogeneous migrations (SQL Server to SQL Server) and heterogeneous ones (Oracle to PostgreSQL) when paired with the Schema Conversion Tool.
- AWS Application Migration Service (MGN) performs lift-and-shift rehosting of entire servers, continuously replicating block-level changes so the cutover window is minutes rather than hours.
- The AWS Snow family (Snowball, Snowball Edge, Snowmobile) physically transports very large data volumes where network transfer would take impractically long.
- AWS DataSync automates online transfer of file data between on-premise storage and S3 or EFS, with built-in verification.
- AWS Migration Hub gives one place to track migration progress across all of these tools rather than checking each console separately.
Connectivity: VPN and Direct Connect
Most migrations need reliable connectivity between the existing data centre and AWS during the transition period, sometimes for months. A Site-to-Site VPN is quick to set up and adequate for smaller data volumes or lower-priority connectivity; AWS Direct Connect provides a dedicated, private network connection with more predictable latency and throughput, and is worth the lead time to provision for a migration of any real size.
Sizing and cost: right-sizing recommendations and Compute Optimizer
AWS Compute Optimizer analyses actual utilisation of running EC2 instances, EBS volumes and Lambda functions and recommends right-sized alternatives — directly addressing the most common source of post-migration overspend, which is sizing new instances against old on-premise specifications rather than real usage. Running this analysis 30 days after cutover, not before, gives it real usage data to work from.
A typical tool sequence, end to end
- Discovery: Application Discovery Service captures utilisation and dependencies across the estate.
- Landing zone: Control Tower and Organizations establish the account structure and guardrails before anything moves.
- Connectivity: Direct Connect (or VPN for smaller estates) is provisioned and tested ahead of the first wave.
- Migration: MGN for server rehosts, DMS for databases, DataSync or Snow family for bulk file and data transfer, tracked centrally in Migration Hub.
- Post-migration: Compute Optimizer and Cost Explorer drive the right-sizing and commitment decisions once real usage data exists.
Frequently asked questions
It is strongly recommended for any estate beyond a handful of servers. Discovery data based on real utilisation and dependencies consistently surfaces workloads and connections that documentation and interviews miss, and skipping it is one of the more common causes of migration surprises.
DMS is worth using whenever downtime needs to be minimised, since it replicates ongoing changes so the cutover is a short switchover rather than a long export-and-import window. For a small, low-change database where an hour of downtime is acceptable, a manual export and import may be simpler.
MGN supports a broad range of source platforms including physical servers, VMware, Hyper-V and other clouds, replicating at the block level so the source operating system and application stack move largely unchanged. Verify the specific source OS and configuration against current AWS documentation before committing to it for a given wave.
The initial setup is typically achievable in days, though customising guardrails, network architecture and account vending to match specific organisational requirements can extend this to a few weeks for a more complex estate.
Not necessarily. A Site-to-Site VPN is often sufficient for a smaller estate or a shorter migration window. Direct Connect earns its setup lead time and cost when data volumes are large, the migration runs over months, or predictable low latency matters for hybrid operation during the transition.
Skipping Compute Optimizer and right-sizing after cutover, and instead leaving instances sized to match the old on-premise hardware indefinitely. This is one of the largest and most avoidable sources of unnecessary AWS spend in the months after a migration.
How we can help
Migration delivery across AWS, Azure or GCP, with a rollback plan for every step.
Why Your Cloud Bill Keeps GrowingRight-sizing and cost control for the months after cutover.