The client, a global industrial equipment manufacturer with 22 production plants and 5 regional data centers across 14 countries, was running its business on more than 800 physical x86 servers — many in plant server rooms, some untouched for a decade — alongside 1,400 virtual machines fragmented across Hyper-V, KVM, and out-of-support vSphere clusters inherited through acquisitions. Three hypervisors meant three toolchains, three backup regimes, and three sets of skills, while the physical estate carried single-server failure risk under production-critical applications. Plant maintenance windows were short and negotiated months ahead, many physical servers ran vendor software certified only on bare metal or tied to USB dongles and serial-port shop-floor connections, and no single inventory of the estate existed.
Aconic standardized the target first — VMware clusters in the regional data centers plus right-sized plant-edge clusters where shop-floor latency demanded local compute — then routed every workload through one of three migration lanes chosen by evidence gathered during discovery. Physical servers moved through a P2V lane using replication-based conversion with a rehearsal boot in an isolated bubble network before any cutover was scheduled, so every migration had already succeeded at least once by cutover day. Hyper-V and KVM workloads converted through HCX OS-assisted migration, and legacy vSphere workloads moved live via HCX vMotion with no downtime window at all.
Aconic's Migrazo platform built the first complete inventory of the physical and virtual estate, capturing utilization, installed software, and peripheral dependencies like USB dongles and serial connections, and gave every asset an explicit disposition: virtualize, convert, retire, or retain. Discovery found 9% of the physical estate was doing nothing at all and retired it without migration. Nine migration waves moved the estate plant-cluster by plant-cluster, sequenced around production calendars, with all three lanes running concurrently and every workload passing automated post-migration validation before its wave closed. Workloads that failed virtualization criteria — dongle-locked software, serial-attached shop-floor equipment, strict vendor support matrices — were never forced: 112 servers were deliberately retained on refreshed bare metal, each with documented rationale.
Over 12 months, the client virtualized or retired 86% of its physical server estate, standardized three hypervisor platforms down to one, and cut rack space by 61% and power draw by 52%. Average P2V cutover downtime came in under 25 minutes with a 99.4% first-attempt success rate across 1,790+ migrations, and 5-year infrastructure TCO dropped 34% — all without stopping a single production line. As the client's VP of IT Infrastructure put it: "The rehearsal boots were what won our plant managers over ... we virtualized machines that had been running untouched since 2014, and production never noticed."