SaaS company: VMware to CloudStack migration

A SaaS company provisioned every VM by hand on VMware. We moved the platform to Apache CloudStack, with templated deployment, a provisioning API and self-service networking, and migrated 30+ VMs with the SLA never breached.

October 2026  ·  3 min read

SECTOR

SaaS company

SCALE

30+ VMs migrated

PLATFORM

VMware vSphere to Apache CloudStack, fibre-channel storage, Veeam

ENGAGEMENT

Migration project with 24/7 L2/L3 engineers on watch and on call

30+

VMs migrated, SLA never breached

< 30 s

New VM ready via API, from at least 6 hours

10–15 min

Downtime per VM, non-peak hours only

Environment

The problem

Every VM was built by hand. Each new VM on VMware took hours of manual work, including the software setup. The overhead grew with every customer, and the platform team could not keep pace with demand.

Approach

Move the platform, not just the VMs. We deployed Apache CloudStack with fibre-channel storage, built current OS templates and added a provisioning API. Existing VMs were migrated in place, with MAC addresses pinned so that no IP or network configuration had to change inside the operating systems.

Templates and an API turned a six-hour job into a seconds-long one.

What changed

No IP changes, SLA never breached. Pinned MAC addresses meant no network changes inside the operating systems. 24/7 L2/L3 engineers watched every operation live.

At a glance

Challenge

A SaaS platform on VMware with manual provisioning and growing overhead.

Approach

CloudStack with templates, a provisioning API and pinned MACs; VMs migrated in place, one at a time.

Result

30+ VMs migrated, SLA never breached. New VMs ready in under 30 seconds. Client now has a control panel for CPU, memory, disk and IPv4 per project.

The migration

  1. Build and test: templates, fibre-channel storage, Veeam and the API
  2. Import VMs: each one shut down and started gracefully in non-peak hours, 10–15 minutes of downtime per VM
  3. Move public-facing VMs: load balancers and web servers moved one at a time

Outcome

Self-service and an API for the client. The client creates its own clusters and VPN endpoints, deploys VMs through the API in under 30 seconds, and watches CPU, memory, disk and IPv4 allocation per project in a control panel. Backups are managed centrally with Veeam.

More case studies

Hypervisor I/O Stalls

40 VMs returned to stable I/O

Mail Reputation Containment

18 h To confirmed blacklist removal

InnoDB Crash Recovery

0 Data loss identified

Want the same result for your platform?

Tell us what you run. We will tell you how we would move it, and support it afterwards.