Tesco moving 40k server workloads off VMware amid Broadcom's abusive conduct

Original Article ↗ Hacker News Discussion ↗

What Tesco Might Be Migrating To

  • Commenters speculate about alternative virtualization platforms: OpenShift Virtualization (KubeVirt-based), Nutanix, Proxmox, OpenStack, and XCP-ng are mentioned.
  • The article’s note that Tesco’s new platform is incompatible with existing backup tools makes some rule out specific products (e.g., Morpheus, Proxmox + Veeam, which are known to integrate).
  • No consensus emerges; the specific replacement is unclear.

Scalability and Technical Fit

  • Several argue Proxmox would be unwieldy at 40k VMs; practical cluster sizes are reported around a couple dozen nodes, implying many clusters.
  • Others counter that multi-cluster Proxmox at that scale is feasible (tens or hundreds of clusters), but possibly operationally painful.
  • OpenStack is cited as having proven large-scale deployments (e.g., CERN), and Nutanix/OpenShift are viewed as more “enterprise-ready” for this size.

Migration Effort and Timelines

  • A practitioner from a vendor side notes that 40k VM migrations are now routine but still large; 500–1000 VMs/day is possible once tooling and planning are in place.
  • Organizational factors (procurement, compliance, legal, coordination between old/new infra teams, user training) are seen as dominating the 18‑month timeline more than raw technical work.
  • Data movement, storage constraints, legacy OSes, weird appliances, and downtime windows (e.g., shutting production lines on weekends) all add complexity.

Broadcom/VMware Business Model

  • Broadcom is widely portrayed as deliberately milking a declining but sticky VMware base: massive price hikes, cost-cutting, aggressive compliance/enforcement.
  • Many report enterprises renewing at “insane” rates because they can’t move quickly; others view this as foreseeable and blame poor long‑term planning.
  • The strategy is likened to private equity: acquire cash-generating products, slash investment, raise prices, accept long-term customer loss.

Vendor Lock-in and Alternatives

  • VMware is described as having been “in a league of its own” and deeply embedded (backups, monitoring, networking, storage, Tanzu/NSX/VSAN), making exits hard.
  • Alternatives (OpenShift Virtualization, Nutanix, Proxmox, KVM on Linux, OpenStack, XCP-ng) are considered increasingly viable, but often with reduced or different functionality compared with full VMware stacks.

Scale of Tesco’s Infrastructure

  • Some are surprised a retailer needs 40k VMs; others point to thousands of stores, per-store clusters, logistics, analytics, and multiple environments (dev/test/stage/DR) as easily justifying that magnitude.
  • Whether “40k workloads” maps to VMs, servers, or something else is noted as somewhat ambiguous.