Live storage migration under production load

60+ production VMs moved to a new SAN backend across a 5-day window, with no maintenance window and zero customer-reported downtime.

5 September 2026  ·  2 min read

SECTOR

Virtualization provider

SCALE

60+ production VMs

PLATFORM

VMware vSphere + shared SAN

ENGAGEMENT

Planned migration

60+

VMs migrated live

5 days

Total migration window

0

Customer-reported downtime

Environment

Staged storage vMotion schedule across five daysMigration ran in controlled batches: high-I/O virtual machines were moved during overnight windows, standard workloads during lower-traffic periods, each batch followed by datastore and snapshot verification.60+ VMs moved live, in batches, over five daysDay 1Day 2Day 3Day 4Day 5High-I/O VMsStandard VMsVerify + scrubshaded = overnight window, used for the highest-I/O VMs60+VMs migrated5 dayselapsed0maintenance windows0reported downtime

Scroll the diagram sideways to see all of it →

Batching by I/O profile is what removed the need for a maintenance window.

The problem

Hardware lifecycle set the deadline; the business set the constraint. The SAN controller backing the primary datastore was approaching end of support. All VM storage had to move to a new backend without customer-visible downtime during business hours.

Investigation

Profile first, migrate second. We assessed the storage layout, identified the VMs with high I/O profiles that needed priority handling, and mapped the snapshot dependencies. Storage vMotion behaviour was then tested under load against a staging subset to validate performance impact thresholds before anything production moved.

The absence of a maintenance window was the real constraint — not the migration itself.

Root cause

Not a fault. This was a planned migration driven by hardware lifecycle. The engineering challenge was executing it live, with no maintenance window available.

At a glance

Challenge

SAN controller approaching end of support, with no maintenance window available.

Approach

Profile I/O per VM, validate vMotion under load in staging, then migrate in batches.

Result

60+ VMs moved over 5 days. Old SAN decommissioned on schedule.

The fix

Controlled batches, weighted by I/O profile. We ran staged storage vMotion during lower-traffic periods, with high-I/O VMs prioritised into overnight windows. Latency and throughput were monitored continuously through each batch, and datastore integrity and snapshot chain health were verified after every migration.

Outcome

All 60+ VMs moved with zero customer-reported downtime. The migration completed over a 5-day window and the old SAN was decommissioned on schedule.

More case studies

WordPress Performance Recovery

1.8 s Page load, down from 12 s

Hypervisor I/O Stalls

40 VMs returned to stable I/O

Mail Reputation Containment

18 h To confirmed blacklist removal

Hardware lifecycle should not cost you a maintenance window

Staged, monitored and reversible. We plan storage migrations around your traffic rather than around a Sunday night outage notice.