/// case_study · aws_migration

A 47-server VMware estate, migrated to AWS: agent-accelerated.

We migrated a Greek enterprise's entire on-premises VMware estate to AWS and modernised its security stack to cloud-native: planned and executed with AWS Transform's agentic AI and a custom migration agent.
/profile
A Greek enterprise · 47-server VMware estate
/approach
Agentic migration with AWS Transform
/where_it_runs
AWS · EU (Frankfurt)
/shape
5 waves · ~12 weeks
/// the_problem

A growing enterprise was running its whole business on an ageing, on-premises VMware estate, and VMware had become more expensive and less certain to stay on.

  • 47 servers, mixed and sprawling: Windows and Linux, spanning business apps, databases, and infrastructure services, some of it end-of-life.
  • A hand-built security stack: a six-server, self-run security operations cluster that the team had to patch, host, and babysit.
  • VMware cost & uncertainty: rising licensing and a strategic need to get off the platform onto something modern.
  • Migration is slow and risky by hand: discovering dependencies, planning waves, and rebuilding the network manually is where migrations stall.

They wanted off VMware and onto AWS, without a year-long project or a big-bang cutover.

/// the_solution

We ran the migration with AWS Transform for VMware, an agentic AI service that discovers the estate, maps dependencies, plans the waves, translates the network to AWS, and rehosts the servers, with our engineers steering and signing off each step.

01

Discover

Agentic discovery of the VMware estate and a full dependency map: no guesswork about what talks to what.

02

Plan waves

An AI-generated migration plan: applications grouped and sequenced into five waves, an 8-server pilot first.

03

Landing zone & network

A Control Tower landing zone, and the VMware network translated to an Amazon VPC, with a Site-to-Site VPN for the migration.

04

Migrate

Iterative, wave-by-wave rehosting onto right-sized EC2. Both environments live until each cutover is signed off.

// AWS reports AWS Transform can cut migration execution time by up to 90% versus traditional methods. Human-in-the-loop throughout: nothing cuts over without engineer and client sign-off.

/// by_the_numbers
47
servers migrated to AWS, the on-prem VMware estate retired
~90%
faster migration execution with AWS Transform (AWS benchmark)
~$20k
lower run-rate per year with cloud-native security
6→0
self-run security servers, replaced by AWS-native services

// server count and waves from the delivered plan; cost figures modelled (On-Demand, Frankfurt); the 90% figure is AWS's published benchmark for AWS Transform.

“An entire VMware estate moved to AWS, with the discovery, wave planning, and network design done by agentic AI, and a custom agent for the parts the presets don't cover.”
/// what_we_delivered
01

Agentic discovery & wave planning

AWS Transform for VMware

AWS Transform discovered the estate (via discovery collectors and the Export for vCenter tool), mapped every dependency, and generated a migration plan that grouped applications and sequenced them into waves, with an 8-server Linux pilot to validate the process before higher-risk workloads moved. Databases were isolated into their own wave with coordinated downtime.

02

AI network translation & landing zone

VMware → VPC

The service translated the on-premises VMware network into an Amazon VPC architecture, and we stood up an AWS Control Tower landing zone (multi-account, least-privilege IAM, MFA) plus a Site-to-Site VPN that kept both environments connected throughout the migration.

03

A custom migration agent

agent builder toolkit

Beyond the presets, we built a custom transformation agent with AWS Transform's agent builder toolkit, encoding this estate's specifics: mapping the self-built six-server security cluster onto AWS-native equivalents, flagging the two end-of-life servers for as-is rehosting with explicit risk notes, and applying the client's tiered, deny-all security-group baseline consistently. The agent improves from each wave's results.

04

Cloud-native security modernisation

retire the SOC cluster

Rather than rehost the six-server security stack on EC2, we replaced it with AWS-native services: GuardDuty, Security Hub, AWS Network Firewall, Amazon Detective, AWS Config, and Inspector v2, so the client gets managed threat detection and compliance with nothing to patch or host, plus AWS Backup across every instance.

/// the_modernisation_choice

Two routes were modelled: rehost everything, or modernise security on the way in. They chose to modernise.

Rehost everything: migrate all 47 servers to EC2, including the six-server security cluster running inline, a faithful copy of on-prem, but still a stack the team has to operate. Modelled at roughly $8,500 / month of AWS infrastructure.

Modernise security (chosen): migrate the workloads, but replace the security cluster and select infrastructure with AWS-native services. Fewer servers to run, managed security, and a lower bill, modelled at roughly $6,800 / month, about $20k/year less, while removing six servers' worth of operational burden.

// modelled AWS estimates, On-Demand, Frankfurt. Committed-use pricing would reduce EC2 compute a further 30–40% after a stabilisation period.

/// how_it_works
the agentic engine

AWS Transform for VMware

A generative-AI service (large language models + graph neural networks) that automates the full lifecycle: discovery, dependency mapping, network translation, wave planning, and EC2 rehosting, with human-in-the-loop controls.

the bespoke part

Custom agents

AWS Transform's agent builder toolkit lets us build agents tailored to a client (their security model, their edge cases, their standards) that plug into the same agentic workflow and learn from each run.

the AWS foundation

Landing zone + EC2

AWS Control Tower governance, right-sized EC2 across two Availability Zones, AWS Backup on every instance, and cloud-native security, all defined and deployed as repeatable infrastructure.

services involved: aws

/// why_it_works

Off VMware, onto AWS: faster, and with less to run.

01The whole 47-server estate on AWS, on-prem VMware retired
02Agentic migration: discovery, wave planning, and network design done by AI
03A custom agent for the bespoke work the presets don't cover
04Cloud-native security: six self-run servers gone, ~$20k/yr lower run-rate
/// ready_when_you_are

Planning a VMware exit or an AWS migration?