Euro-cloud provider Anexia moves 12,000 VMs off VMware to homebrew KVM platform

Original Article ↗ Hacker News Discussion ↗

A European cloud provider has migrated 12,000 virtual machines from VMware to a homegrown KVM/QEMU-based platform, highlighting how Broadcom’s aggressive price hikes and contract terms are pushing customers to abandon VMware. Commenters note that while 12,000 VMs is significant but not hyperscale, the real challenge is operational — tooling, integrations, certification requirements, and customer coordination — rather than hypervisor technology itself. Many expect growing demand for mature KVM-based alternatives (Proxmox, OpenStack, Red Hat solutions, etc.), and see Broadcom’s strategy as destroying long-term value by driving even large, previously “captive” enterprises to plan their exits.

Scale and Significance of 12,000 VMs

  • Commenters generally see 12k VMs as “medium to large” but not extreme in VMware terms; some single enterprises run ~30k+.
  • Impact depends heavily on context: VM size, workload density, and number of customers behind them.
  • 12k VMs could represent full stacks for thousands of SMEs or a substantial portion of a country’s healthcare infrastructure.
  • Hardware estimates vary widely: from a single dense rack to many racks, depending on utilization and power per rack.

Migration Complexity and Process

  • Difficulty is less about technology and more about people, documentation, and ownership.
  • Single-tenant vs multi-tenant:
    • Single large customer: easier coordination but messy internal estates, missing docs, unknown app owners.
    • Cloud provider: has clear customer contacts but less control over guest OS and must persuade customers to cooperate.
  • Example Red Hat consulting experience: 12k-VM conversions take 1–3 months, with per-VM downtime from a reboot to many hours.
  • Anexia’s case was eased by:
    • Existing KVM-based know‑how via another hosting brand.
    • Data already on NetApp, independent of VMware storage.

KVM, “Homebrew Platform,” and Tooling

  • “Homebrew KVM platform” is read as a KVM+QEMU+libvirt stack with custom orchestration, not raw KVM alone.
  • Many note that when people say “KVM” they usually mean the full libvirt/QEMU ecosystem.
  • VMware’s main advantage has been integrated networking, storage, and management; replicating this with OSS is doable but staff-intensive.

Alternatives to VMware

  • Options discussed: Proxmox (improving but not VMware‑level for large deployments), OpenNebula, OpenStack, KubeVirt, oVirt, and generic KVM.
  • Some workloads that once used VMware now run on Kubernetes; others still require certified VMware environments (e.g., large enterprise apps/healthcare).

Broadcom Strategy, Lock‑in, and Customer Reactions

  • Broadcom’s price hikes and shift to large upfront payments are widely seen as rent extraction based on lock‑in.
  • Views differ on whether this is rational monetization or value‑destroying short‑termism.
  • Many report organizations either actively migrating off VMware or planning to, though some large or highly entangled environments will likely just “pay up.”
  • Tactics like huge renewal quotes are described as forcing C‑level attention, sometimes backfiring when customers choose to exit.