VergeOS

Two controller nodes for a production system (N+1); three for N+2. A single-node host is documented for homelab/test only.

200 nodes and 100 clusters per system; multiple systems managed from the Sites dashboard.

Single-node VM host shown in the homelab reference architecture for test; no single-node production redundancy.

Two-node HCI is the minimum (edge reference design with directly connected nodes); no external witness required. N+2 recommends 5 nodes to avoid split-brain.

Not supported: all nodes must share one switching fabric with zero switch hops (under 0.05 ms); use site sync for a second site.

Yes - compute-only, storage-only and scale-out nodes; storage tiers span clusters in one system.

Fibre Channel LUNs as vSAN tier disks (one LUN per node), NFS/CIFS remote volumes via NAS, or in-guest iSCSI.

NVMe direct-attached preferred; Tier 0 NVMe for metadata; all-NVMe tiers supported.

Global inline deduplication across the whole system (production, snapshots and replication), plus compression; included.

No erasure coding; replication factor RF2 (default) or RF3.

N+1 (RF2, default) or N+2 (RF3, from 26.1); per-tier redundancy, 2 vSAN fault domains; ioGuardian repair server can refill missing blocks beyond RF.

ioClone: instant, space-efficient, independent snapshots/clones of VMs, volumes, tenants or the whole system; scheduled snapshot profiles.

AES-256 vSAN encryption chosen at install; keys on USB drives or a startup password. No external KMS. Site syncs encrypted by default.

Tier placement control (preferred tier, live moves between tiers); no documented IOPS/throughput limits.

Built-in NAS service with NFS and CIFS/SMB shares, AD join and volume snapshots.

No native S3 object service documented.

Not documented; storage is consumed by VergeOS VMs, tenants and NAS.

Site syncs: block-level, deduplicated, compressed, AES-256 replication of full or tag-based partial snapshots; intervals as short as minutes (5-minute example in 26.1); many-to-one.

No synchronous replication between sites.

ioReplicate: three-click recovery of whole Virtual Data Centers, non-disruptive DR tests from snapshots, BGP-based network failover.

Built in: system/tenant/VM snapshots, off-site site syncs and ioGuardian; a Backup licence covers dedicated backup/DR target systems.

Immutable system snapshots (no early deletion, up to 7-day unlock delay with alerts) and ioFortify ransomware detection alerts.

Veeam Backup & Replication 13.1+ native integration (announced Aug 2026, incl. per-tenant); Storware also integrated.

Replicate to a VergeOS system on bare metal at a hyperscaler, colo or service provider; no native AWS/Azure service.

VergeFabric: internal L2/L3 networks, routing, NAT, DHCP/DNS, BGP, WireGuard and IPsec VPNs, per-network firewall; included.

Per-network firewall rules, isolated internal networks and tenant isolation; no per-VM distributed firewall policy engine.

Users, groups and permissions; OIDC SSO (Entra ID, Google, others); TOTP two-factor; API keys.

Own boot-image integrity check instead of UEFI Secure Boot on nodes; guest vTPM; no STIG or drift remediation published.

No FIPS 140, Common Criteria or STIG attestations published.

Up to 100 clusters in one system plus the Sites dashboard and global/central remote management across systems.

One software package; minor/patch updates apply live without workload impact; node firmware handled separately.

REST API, Terraform provider, Ansible, PowerShell module, Python/Go/TypeScript SDKs, vrg CLI, webhooks and an MCP server.

Dashboards, alarms, diagnostics, Prometheus exporter and tenant usage reports; no predictive capacity planning documented.

Nested tenants (Virtual Data Centers) with their own UI, networks and quotas; VM and tenant recipes, marketplace, usage reports for chargeback.

Live migration, automatic HA restart and automatic evacuation during maintenance; ioOptimize moves workloads off failing hardware.

CSI driver (block RWO and NFS RWX) and Cloud Controller Manager for any Kubernetes on VergeOS VMs; Rancher integration.

VergeOS can run on bare metal at hyperscalers or colos; no managed public-cloud service or cloud control plane.

Embedded VDI desktop plus Inuvika and Leostream integrations; no Citrix or Omnissa validation published.

GPU/PCI passthrough, NVIDIA vGPU and MIG (NVIDIA licence needed); VergeIQ private AI with OpenAI-compatible API included since VergeOS 26.

None (recipes can template database VMs).

Boot ISO/USB installer on each node (controllers first, then scale-out nodes); a system can run within hours.

Any x86 server meeting sizing rules (enterprise SSD/NVMe, JBOD/HBA, Intel/NVIDIA Mellanox/Broadcom NICs), refurbished hardware, partner-certified appliances.

Two-node edge clusters on small-form-factor hardware with central Sites dashboard and backup to a core system.

Supported with an air-gap licence file requested by email (requires justification) and offline updates.

ioMigrate connects to vCenter and uses VMware backup APIs with CBT for near-zero cutover; imports of VMDK/VHDX/OVA/QCOW2; Cirrus Data integration.

No free edition; 14-day hosted test drive, proof-of-concept and trial/NFR licences.

Licence follows the system ID across hardware changes and vendors; VergeOS can also run on bare-metal cloud servers.

One vendor for the whole software stack; hardware support stays with the server vendor or appliance partner.

No published response SLA; reseller SKUs are labelled "24x7 4hr included".

Small but growing: Veeam, NVIDIA, Intel, Storware, Leostream, Inuvika, Cirrus Data; distribution via TD SYNNEX and Ingram; Discord community.

Standard KVM guests with VM export, but storage, networking and management are proprietary; exit means VM migration.