# Fabric by IZT Cloud — full product reference

> HCI as firmware. Not a license you rent by the core.

Last updated 2026-09-03. Canonical site: https://fabric.izt.cloud/. Short index: https://fabric.izt.cloud/llms.txt.

Fabric is a hyperconverged infrastructure (HCI) platform from IZT Cloud that ships as one signed firmware image per server: a type-1 hypervisor, distributed block storage, per-tenant software-defined networking, high availability with verified fencing, a tenant self-service portal and signed A/B updates — on commodity x86 hardware, priced per node by node class, never per core, and fully operable with no internet connection.

**Release status:** Fabric is in closed beta today. General availability is planned for Q1 2027.

Fabric is built and sold by IZT Cloud (InZernet Technologies, Inc.). It is a hyperconverged infrastructure (HCI) platform and a VMware alternative for organisations leaving per-core licensing after the Broadcom acquisition, for edge and branch sites, for sovereign and air-gapped deployments, and for managed service providers who need multitenancy included.


## Vocabulary

- **Fabric Compute** — the hardware-accelerated type-1 hypervisor layer built into the node image.
- **Fabric Storage** — the distributed block storage pooled across every node, driven by storage policies.
- **A fabric** — one site: one cluster, one LAN, one control-plane appliance. A second datacenter is a second fabric.
- **Node class** — the licensing band a server falls in by physical core count (Edge, Small, Medium, Large, X-Large).
- **Tenant** — an isolated customer or department with quotas, its own virtual networks and a self-service portal.
- Health is reported as Healthy / Degraded / Critical; disks are disks, control nodes are control nodes.

## Pages

- [Fabric — HCI as firmware, not a license you rent by the core](https://fabric.izt.cloud/) — Fabric is a hyperconverged infrastructure platform delivered as one signed firmware image: compute, distributed storage, networking, HA and a tenant portal on commodity servers, priced per node, never per core.
- [Platform — One signed image, the whole stack](https://fabric.izt.cloud/platform) — Every layer of the Fabric HCI platform — Fabric Compute, Fabric Storage, per-tenant SDN, HA with verified fencing, signed A/B updates and air-gapped operation — in one firmware image.
- [How Fabric scales — from one edge node to a 500-node datacenter](https://fabric.izt.cloud/scale) — See a Fabric cluster grow from one node to five hundred and fail over from edge to datacenter, live. Same firmware image at every size, no re-architecture.
- [Use cases — Edge, sovereign, AI/GPU, MSP and VMware migration](https://fabric.izt.cloud/use-cases) — Where Fabric fits: retail and branch edge clusters, air-gapped industrial and public-sector sites, GPU workloads, service providers, and organisations leaving VMware after Broadcom.
- [Pricing — Per node, per year, every feature included](https://fabric.izt.cloud/pricing) — Fabric HCI pricing: five node classes from $500 to $5,000 per node per year by physical core band, every feature in every class, a free unlimited Community edition, and no per-core meter.
- [Compare Fabric with VMware, Proxmox, Nutanix, Scale Computing and more](https://fabric.izt.cloud/compare) — Honest, side-by-side comparisons of Fabric with every major HCI and virtualization platform: licensing model, architecture, storage, networking, HA, updates and sovereignty.
- [About — Infrastructure should belong to the people who run it](https://fabric.izt.cloud/about) — IZT Cloud builds Fabric on one belief: the people who run infrastructure should own it. No per-core meter, no phone-home, no kill-switch.

## Pricing

Unit: per physical node, per year, USD list; subscription only. Every class is the whole platform; there is no feature matrix.

| SKU | Class | Cores per node | $ per node per year | MSP $ per node per month |
|---|---|---|---|---|
| FAB-EDGE | Edge | up to 8 cores | $500 | $50 |
| FAB-S | Small | 9 – 16 cores | $1,000 | $100 |
| FAB-M | Medium | 17 – 32 cores | $1,800 | $180 |
| FAB-L | Large | 33 – 64 cores | $3,200 | $320 |
| FAB-XL | X-Large | 65+ cores | $5,000 | $500 |

- Community — free — the whole platform, unlimited nodes, forever; community update channel and forum support, no SLA; commercial use permitted
- Term: 3-year term −10%/yr · 5-year term −15%/yr (prepaid).
- Fleet: fleet discount on total nodes per organisation: 9–24 −15% · 25–64 −25% · 65–199 −35% · 200+ contact us.
- Standard — business hours, next-business-day, unlimited tickets — included in every paid class (deployments under $2,500/yr: portal + email, 3 tickets/yr).
- Premium 24×7 — +30% — 24×7×365, 1-hour critical-incident response, proactive telemetry.
- Mission Critical — +75% — 24×7×365, 30-minute critical response, named engineer.
- MSP / service provider — monthly per node from $50 (Edge) to $500 (X-Large), no minimum term, multitenancy included.
- OEM / embedded — per-node royalty for hardware partners shipping Fabric pre-installed.

Pricing promises:

- No per-core meter — five wide classes, set once by physical core count; density inside a class is free.
- No renewal shock — multi-year terms with a written cap on renewal increases.
- No feature matrix — every class is the whole platform.
- Switch for less — VMware migration credit through your existing contract’s expiry; education and non-profit discounts.

Worked comparison — One dual-socket server, 32 physical cores, fully loaded with VMs (annual, list unless noted): Fabric Medium class $1,800 per node per year.

- VMware: ~$6,400 – $11,200 / node / yr (VCF at ~$200/core realized to ~$350/core list, 32 cores, 1-year term (industry benchmarks, 2026)). Model: Per core, subscription only (perpetual licences ended in 2023), 16-core minimum per CPU socket, sold as VVF or VCF bundles.
- Proxmox: ~€1,100 (Standard) to ~€2,200 (Premium) / node / yr (two sockets, published subscription prices, 2026). Model: Free (AGPL). Optional subscription per CPU socket for the enterprise repository and support: Community €120, Basic €370, Standard €550, Premium €1,100 per socket per year; all nodes in a cluster on the same level.
- Nutanix: ~$20,800 / node / yr (NCI Pro with Production support at ~$649/core reseller list, 32 cores, 2026). Model: Per physical core, subscription, in Starter / Pro / Ultimate tiers with support level multipliers; Prism Central, Flow and DR features vary by tier or add-on.
- Scale Computing: ~$6,400 / node / yr (verified 5-year street rate of ~$201/core from reseller catalogs and a transacted public-sector contract, 32 cores, August 2026). Model: Per physical core, subscription (1/3/5 years), channel-only; Standard vs Professional tiers, Fleet Manager licensed separately per cluster.
- VergeIO: ~$12,000 / node / yr (VergeOS Enterprise per-node reseller pricing, 2026 (a 5-year Enterprise term lists at ~$61,760 per node)). Model: Per physical server, subscription, all features included; SMB bundles with VM caps and Enterprise per-node terms.
- Harvester: Free software; support priced by SUSE subscription (public SUSE pricing is quote-based, 2026). Model: Free and open source (Apache 2.0). Enterprise support through SUSE subscriptions, typically alongside Rancher Prime.
- Azure Local: Per-core monthly Azure billing plus guest licences (Microsoft publishes regional per-core rates; totals depend on Azure agreement and guest OS licensing, 2026). Model: Azure Local: per physical core per month, billed through an Azure subscription, plus Windows Server guest licensing. Hyper-V: included with Windows Server Datacenter, per core with a 16-core minimum per server.
- XCP-ng: Free software; support per host via Vates (Vates publishes tiered plans; totals depend on host count and plan, 2026). Model: Free hypervisor (GPL). Vates support per host and Xen Orchestra Appliance subscription plans; hyperconverged storage (XOSTOR) supported as an add-on.

## Pricing FAQ

**Why five classes instead of one flat price?**

Because one flat price is either too expensive for a four-core edge box or too cheap for a 128-core enterprise host. A class is a wide band, set once by the node’s core count, and inside a band density is free: moving from 20 cores to 32 costs nothing. We classify nodes; we don’t meter them.

**What happens if I put a bigger CPU in an existing node?**

If it stays inside its class, nothing changes. If it crosses into the next class, your portal shows the new class and the price difference, and you request an updated entitlement token. Replacing a failed node with an identical one never needs anything from us.

**How is licensing enforced?**

Each fabric holds a signed token stating how many nodes of each class you have. Your nodes verify it locally, offline. There is no license server in your cluster, no phone-home, and nothing we can switch off remotely — the subscription grants access to the signed production update channel.

**Why no perpetual license?**

A hypervisor must stay patched against new vulnerabilities. Perpetual licenses encourage running unpatched, internet-facing infrastructure, a security risk we won’t design in. Subscription keeps every node on the signed security-update channel.

**Can I use my existing hardware?**

Yes. Fabric installs on commodity x86 servers, no restrictive hardware compatibility list, with growable RAID-1 from one disk to many.

**Can we run completely disconnected?**

Yes — it is designed for it. There is no phone-home and no licence server in your cluster. Sign-in is local (password, TOTP or passkey, never an emailed link), your control plane runs on your own appliance, and firmware arrives as a signed image you import into the fabric’s own library and verify before it rolls. Nothing in the platform needs to reach the internet.


## Features

Status is stated per feature. "Shipping" is in the current firmware and control plane; "Roadmap" is planned and not yet available.


### Fabric Compute

A hardware-accelerated type-1 hypervisor built into the signed image, driven by one control plane across every node.

- **Full VM lifecycle** (shipping) — Create, start, stop, reboot, resize, clone-from-ISO, attach media and delete VMs from the web portal or the REST API. Every mutation is an auditable operation.
- **Live migration** (shipping) — Move a running VM between nodes with no downtime. Storage is already shared cluster-wide, so only memory moves — and the target pre-builds the tenant network before the guest arrives, so a move never fails for a network the target had not seen.
- **Cold migration and hibernate** (shipping) — Park a VM’s RAM to distributed storage and resume it on any node, or move a stopped VM to a different node in one action. Resizes never refuse; they apply at the next power cycle, and the control plane re-places the VM where it fits.
- **High availability with verified fencing** (shipping) — Ten-second acknowledged heartbeats, a liveness verdict in about twenty seconds, and a hard rule: no VM is restarted elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. Split-brain is designed out, not merely discouraged.
- **Guest memory admission control** (shipping) — A VM cannot be placed, started or adopted onto a node that lacks the RAM for it, and a runaway guest is contained by per-VM memory limits so one tenant can never take a node down.
- **Maintenance mode with evacuation** (shipping) — Drain a node by live migration, patch it, and return it — the same path the rolling firmware update uses, with a placement check that refuses to start an evacuation the fleet cannot absorb.
- **Compute tiers and policies** (shipping) — Nodes are tiered automatically by CPU model; a compute policy chooses tiers (strict or preferred), CPU exposure and vCPU ratio, so a tenant’s VM lands on the class of hardware you promised.
- **GPU passthrough** (roadmap) — Assign physical GPUs to VMs for AI/ML and graphics with cluster-aware placement. No per-GPU software tax.


### Fabric Storage

Fault-tolerant distributed block storage pooled across every node. No SAN, no separate storage licence, no storage tier to buy.

- **Storage policies** (shipping) — A storage policy is the storage SKU you sell: an ordered list of device tiers (strict or preferred), the number of copies and the minimum to keep serving, a failure domain of host or rack, and per-disk IOPS and bandwidth QoS. Every disk names its policy; the platform builds and maintains the pools behind it.
- **Rack-aware failure domains** (shipping) — Once three racks carry disks the platform switches replica placement to rack-level automatically and tells the operator, so a whole rack can fail without losing data.
- **Inline compression and re-pack** (shipping) — Choose none, fast or dense compression per policy. Because compression applies on write, a one-click re-pack rewrites existing volumes in place with progress, and the policy card shows bytes saved.
- **Live tier moves and online resize** (shipping) — Move a disk between tiers or policies while its VM runs; grow a disk online. Long moves run detached with progress and survive a browser closing.
- **Repair ETA and speedy repair** (shipping) — The storage dashboard shows how long a rebuild will take and lets an operator trade client latency for repair speed, then reverts automatically once the cluster is clean.
- **Physical drive health** (shipping) — Every physical drive — boot mirror and empty bays included — is monitored on the node itself; a failing drive lights its own bay LED even if the control plane is unreachable, and the control plane adds trend analysis so a growing error count is caught before the drive dies. Nothing auto-drains: data movement stays an operator decision.
- **Grow without downtime** (shipping) — Add disks or nodes and the storage layer rebalances automatically; storage telemetry streams into a six-tab dashboard.
- **Instant snapshots** (roadmap) — Native copy-on-write VM and volume snapshots for change windows and backup.
- **Cross-site replication** (roadmap) — Asynchronous per-volume replication to a second fabric with one-command failover. No separate DR product.
- **Kubernetes co-residency** (roadmap) — Expose Fabric Storage policies to Kubernetes through a standard CSI driver and run VMs and containers side by side.


### Networking

Per-tenant software-defined networking that is part of the platform, not a separately licensed SDN product.

- **Tenant virtual networks** (shipping) — Each tenant gets isolated L2 domains carried as an overlay across the fabric, or trunked straight onto an existing VLAN when a network must bridge the customer’s own LAN. Dual-stack IPv4/IPv6.
- **Built-in DHCP and DNS per tenant** (shipping) — Every node answers DHCP and DNS for the guests it hosts, with per-tenant upstream resolvers and automatic guest hostnames — stateless, so there is no DHCP server to fail over. Duplicate-address and rogue-router detection raise events.
- **Micro-segmentation** (shipping) — Named firewall policies enforced at the tenant boundary keep east–west traffic where you put it.
- **Stateless gateways, stateful edges** (shipping) — A node is a gateway because it is cabled to an uplink, never because it was elected. Every VM’s NAT state lives on its own node, so gateways scale out with no state sync and asymmetric carriers are free.
- **ARP or BGP announcement** (shipping) — Public prefixes are announced with plain ARP for simple uplinks or with BGP + BFD + ECMP when the carrier expects a peer — one aggregate, nothing else, enforced by prefix-list.
- **Port groups and cabling discovery** (shipping) — Name your L2 domains, and let the fabric measure which node ports share one — it proposes the grouping and the exact changes, and writes nothing until an operator commits.
- **Link aggregation** (shipping) — Bond a second port with LACP that degrades gracefully to active-backup, so a rolling update is never order-dependent.


### Multitenancy and self-service

Two portals from one build: the operator’s fleet view and a tenant self-service portal, both offline-capable.

- **Tenant self-service portal** (shipping) — Tenants create VMs, disks, networks and ISOs inside their quota, open a browser console or VNC session, and never see another tenant’s resources.
- **Operator fleet console** (shipping) — Discovered nodes, racks, storage policies, firmware rollouts, HA advisories and every long-running operation with progress, in one web console.
- **Browser consoles and remote shell** (shipping) — Serial and VNC consoles for VMs and a root shell or one-shot command on any node — all spliced through a relay the node dials outbound, so nodes run no SSH and expose no inbound ports.
- **Offline authentication** (shipping) — Password plus optional TOTP, or passkeys — never an emailed link, so sign-in works with no internet. Operators finalize new users through single-use enrollment tokens; break-glass is a console command on the appliance.
- **REST API** (shipping) — Everything the portals do is a documented REST call with a uniform envelope; long operations return a pollable Operation so automation never blocks on a slow node.


### Lifecycle and operations

One signed image per node. One rolling update for the whole stack. Automatic rollback when it does not come back.

- **Signed A/B firmware updates** (shipping) — Two image slots on a mirrored boot device. An update stages into the idle slot, verifies against a pinned signing root, flips, and a watchdog rolls back automatically if the node does not come up healthy. Anti-rollback stops downgrade attacks.
- **Rolling fleet update** (shipping) — One click stages every node in parallel, then updates them one at a time: park, evacuate by live migration, flash, reboot, verify, return. A storage-health gate pauses between nodes and the rollout can be cancelled mid-flight.
- **Firmware and ISO libraries** (shipping) — Signed images and install media live in libraries on storage policies you choose; uploads are resumable and verified before import.
- **Self-healing control roles** (shipping) — If a control node stays down, the fabric removes it and promotes a rack-diverse peer automatically, so quorum survives an unattended outage.
- **Node video console** (shipping) — A permanent status screen on every node’s console shows health, mode and a countdown to automatic return — the node asks the fabric what the fabric thinks of it, so a fenced node never claims to be READY.
- **Out-of-band BMC control** (shipping) — Power, identify, boot device and event log through IPMI, driven from the portal. A node manages its own BMC in-band with no credentials; out-of-band actions are relayed through a live peer, so the control plane never needs to reach the management network.
- **Metrics endpoint** (shipping) — One-second VM telemetry, sixty-second guest inventory, storage samples and a Prometheus scrape endpoint — no monitoring licence.
- **Turnkey VMware import** (roadmap) — Connect to an ESXi host, inventory the VMs and migrate them in with automated disk conversion.


### Security and sovereignty

Bring your own hardware, run with no internet at all, and never wait for a vendor to approve your own infrastructure.

- **Immutable, read-only node OS** (shipping) — A read-only root, mirrored boot, signed images and a boot watchdog. There is nothing on a node to drift.
- **Fully air-gapped operation** (shipping) — No phone-home, no licence server in the cluster, no kill-switch. Sign-in is local, the control plane is your appliance, and updates arrive as signed images you import and verify on site.
- **Nodes expose no inbound ports** (shipping) — Nodes connect outbound to the control plane’s message bus and console relay. There is no SSH daemon and no management port to scan.
- **Commodity hardware, no restrictive HCL** (shipping) — Standard x86-64 servers with any mix of SAS, SATA or NVMe drives, growable from one boot disk to a mirror. Mixed node sizes in one fabric are fine — each is licensed at its own class.
- **Product vocabulary in plain language** (shipping) — Health is Healthy / Degraded / Critical; disks are disks and control nodes are control nodes. You never need to learn the internals to run it.


## Use cases

- **Edge and distributed sites** — Two- and three-node HA clusters in stores, branches and remote sites, from $500 per node per year.
- **Industrial and on-prem OT** — Air-gapped clusters on the factory floor with no phone-home and no kill-switch.
- **Public sector and sovereign** — Runs with no internet connection: local sign-in, an on-site control plane, signed updates you import yourself.
- **AI / GPU workloads** — Dense accelerator boxes without a per-GPU or per-core software tax.
- **VMware migration** — Per-node pricing, a migration credit through your existing contract, and a turnkey import path on the roadmap.
- **Service providers** — Monthly per-node billing, no minimum term, multitenancy and a tenant portal included.

## Comparisons

Each comparison scores a fixed set of dimensions. Fabric roadmap items are labelled and scored for the other platform until they ship. Competitor facts come from public documentation; prices carry their basis and date.


### Fabric vs VMware (VMware vSphere and VMware Cloud Foundation (VCF), Broadcom)

URL: https://fabric.izt.cloud/compare/fabric-vs-vmware

Broadcom turned VMware into a per-core subscription sold in a handful of bundles. Fabric is the opposite bet: the whole HCI stack as one signed image, priced per server, with every feature in every class. Here is how they actually compare.

- VMware Cloud Foundation is licensed per core with a 16-core minimum per CPU and bundles vSphere, vSAN, NSX and Aria whether you use them or not. Fabric is licensed per node by class and includes the whole platform.
- VMware remains the deepest ecosystem in virtualization — ISV certifications, backup integrations and talent. Fabric trades ecosystem breadth for one image, one update and one bill.
- On a 32-core node the gap is roughly 3.5–6× per year before discounts; it widens with denser CPUs, because Fabric moves one band and VMware meters every core.

Pricing: VMware — Per core, subscription only (perpetual licences ended in 2023), 16-core minimum per CPU socket, sold as VVF or VCF bundles. Example: ~$6,400 – $11,200 / node / yr (VCF at ~$200/core realized to ~$350/core list, 32 cores, 1-year term (industry benchmarks, 2026)). Fabric — Fabric Medium class: $1,800 per node per year, everything included; 3-year term $1,620/yr.

| Dimension | Fabric | VMware | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Per core, 16-core minimum per socket, subscription only. VCF and VVF bundles; per-core vSAN capacity entitlement (extra TiB is a separate add-on). | Fabric |
| What you install | One signed firmware image on each server and one control-plane appliance you host. Nothing else to license, version or reconcile. | ESXi on each host plus vCenter Server, vSphere Lifecycle Manager, NSX Manager cluster, Aria Suite and SDDC Manager for a VCF stack — each versioned and upgraded separately. | Fabric |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | vSAN (ESA/OSA) with Storage Policy-Based Management — mature and fast, but licensed per core with a capacity entitlement, and only inside VVF/VCF. | Even |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | NSX: the most complete enterprise SDN available, with distributed firewall, load balancing and gateways — as a separately deployed manager cluster inside VCF. | VMware |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | vSphere HA with admission control, mature and battle-tested; fencing relies on datastore heartbeats and isolation response rather than verified out-of-band power-off. | Even |
| Live migration | Live migration between nodes with no downtime; the target pre-builds the tenant network before the guest arrives. Hibernate to storage and resume anywhere. | vMotion, Storage vMotion and cross-vCenter vMotion — the reference implementation. | VMware |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | vSphere Lifecycle Manager images per cluster, with vCenter, NSX and Aria upgraded on their own tracks and an interoperability matrix to check. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | vCenter roles and resource pools; true tenant self-service requires VMware Cloud Director, a separate product. | Fabric |
| Snapshots and DR | (Roadmap) Instant snapshots and asynchronous cross-site replication with one-command failover are on the roadmap, never as a paid add-on. | Snapshots built in; vSphere Replication and Live Site Recovery (formerly SRM) for DR orchestration — a separate licence. | VMware |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Runs disconnected, but the subscription model is built around connected entitlement and usage reporting, and the management stack is large to keep patched offline. | Fabric |
| Hardware | Commodity x86-64 servers, any mix of SAS/SATA/NVMe, mixed node sizes in one fabric. No restrictive compatibility list. | Strict hardware compatibility list; vSAN ReadyNodes for supported storage configurations. | Fabric |
| Edge and small sites | Two- and three-node HA clusters from $500 per node per year in the Edge class. | VCF Edge still requires vCenter; the SMB Essentials kits were discontinued after the acquisition. | Fabric |
| Ecosystem and maturity | New platform with a REST API, Prometheus metrics and a tenant portal; backup-ISV interfaces on the roadmap. | Two decades of ISV certifications, backup vendors, monitoring tools and trained staff. | VMware |
| Migration path | (Roadmap) Turnkey ESXi import (inventory, disk conversion, cut-over) is on the roadmap; today VMs move via standard disk-image import and a migration credit covers your remaining VMware term. | You are already there. | VMware |

Fabric is the better fit for: Organisations facing a Broadcom renewal who want a flat, per-node bill; Edge, branch and mid-market sites Broadcom no longer prices for; Sovereign and air-gapped deployments that cannot depend on connected entitlement; Service providers who need tenancy without a separate cloud-director product.

VMware is the better fit for: Estates with deep NSX, SRM or Aria dependencies; Workloads certified only on vSphere by their ISV; Teams whose entire tooling and staffing is built around vCenter.

Migration: Most VMware exits are a two-step: stand Fabric up beside the existing cluster on commodity or reused hosts, then move workloads tier by tier. Because Fabric is licensed per node rather than per core, the pilot cluster costs the same whether it is three 16-core or three 32-core boxes inside a class, and the VMware migration credit covers you until your current contract expires.

**Is Fabric a drop-in replacement for VMware vSphere?**

It replaces the stack — hypervisor, shared storage, networking, HA and management — rather than any one VMware product. VMs move as disk images; a turnkey ESXi import tool is on the roadmap. Expect a migration project, not a licence swap.

**How does Fabric pricing compare with VMware Cloud Foundation per-core pricing?**

Fabric charges per node by class. A 32-core node is Medium class at $1,800 per year with everything included; VCF on the same node runs roughly $6,400 to $11,200 per year depending on realized versus list per-core rates. The gap grows with denser CPUs.

**Does Fabric include storage and networking, or are they add-ons like vSAN and NSX?**

Both are part of the platform in every class, including the free Community edition. There is no separately licensed storage or SDN product.

**Can Fabric run without an internet connection like an air-gapped vSphere cluster?**

Yes, by design. Sign-in is local, the control plane is your appliance, and updates are signed images you import and verify on site. Nothing phones home and there is no kill-switch.

**What does VMware still do better than Fabric?**

Ecosystem depth: NSX is the most complete enterprise SDN, vMotion is the reference live-migration implementation, and the ISV, backup and DR (Live Site Recovery) ecosystem is unmatched. Fabric’s snapshots and cross-site replication are roadmap items.


### Fabric vs Proxmox (Proxmox Virtual Environment (Proxmox VE), Proxmox Server Solutions GmbH)

URL: https://fabric.izt.cloud/compare/fabric-vs-proxmox

Proxmox VE is the open-source workhorse of the post-VMware world, and it is genuinely cheaper on licence. Fabric is what you buy when you want the same freedom from per-core metering but as a supported appliance rather than a stack you assemble and operate yourself.

- Proxmox VE is free software; the subscription buys the enterprise repository and a ticket allowance, per CPU socket. Fabric’s Community edition is also free and unlimited; paid classes add the production channel and support, per node.
- Proxmox is a general-purpose Debian system you compose: storage, networking, backup and fencing are yours to design. Fabric is a read-only signed image with those decisions made — storage policies, per-tenant SDN, verified fencing.
- Proxmox wins on snapshots, its backup server and the size of its community; Fabric wins on multitenancy, fleet-wide atomic updates and being a product rather than a platform kit.

Pricing: Proxmox — Free (AGPL). Optional subscription per CPU socket for the enterprise repository and support: Community €120, Basic €370, Standard €550, Premium €1,100 per socket per year; all nodes in a cluster on the same level. Example: ~€1,100 (Standard) to ~€2,200 (Premium) / node / yr (two sockets, published subscription prices, 2026). Fabric — Fabric Medium class: $1,800 per node per year including business-hours support; Community is free. Proxmox is cheaper on licence at the Standard tier; Fabric’s value is the integrated appliance and the tenant portal, not a lower list price.

| Dimension | Fabric | Proxmox | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Free and open source. Subscriptions per CPU socket, with cores uncounted; support tiers differ in ticket count and response time. | Even |
| What you install | A signed, read-only firmware image per server plus one control-plane appliance. Every node is identical by construction. | A Debian-based OS per host that you configure and keep consistent; the cluster is formed from the hosts themselves with no separate manager. | Fabric |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | Bring your own: ZFS, LVM, NFS/iSCSI or its integrated distributed storage, each configured and tuned by the operator. Flexible, but there is no policy layer that tenants consume. | Fabric |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | Linux bridges and VLANs by default; an SDN feature adds VXLAN/EVPN zones. Firewalling per VM is available; tenant DHCP/DNS and gateways are operator-assembled. | Fabric |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | HA manager restarts VMs after a node is lost, with self-fencing via watchdog. No out-of-band power verification by default, so shared-storage safety depends on the watchdog firing. | Fabric |
| Live migration | Live migration with pre-built tenant networking on the target; hibernate and resume on any node. | Live migration with or without shared storage, including local-disk migration. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Package updates per node from the enterprise repository, applied and rebooted by the operator; no image-level rollback if a node fails to come back. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Users, groups, pools and fine-grained permissions on one management UI — enough for departments, not a tenant portal with isolated networks per customer. | Fabric |
| Snapshots, backup and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap. Third-party backup integration is planned through standard interfaces. | VM snapshots, scheduled replication between nodes, and Proxmox Backup Server with deduplication and verification — a real strength. | Proxmox |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Runs fully offline; you mirror the repositories yourself. Subscription keys validate against Proxmox servers unless offline-activated. | Even |
| Hardware | Commodity x86-64, any drive mix, no restrictive compatibility list. | Anything Debian runs on — the broadest hardware support of any platform here. | Even |
| Support | Business-hours support included in every paid class; 24×7 with 1-hour critical response at +30%; named engineer at +75%. | Business-hours only (CET) at every tier; ticket-capped below Premium. 24×7 comes from third-party partners. | Fabric |
| Ecosystem and community | Young product; REST API and Prometheus metrics; backup-ISV interfaces on the roadmap. | Very large community, extensive documentation, Terraform and Ansible providers, years of production use. | Proxmox |
| Operational skill required | Plain-language health (Healthy / Degraded / Critical), disks and control nodes; you do not learn the internals to run it. | You are the storage, network and fencing architect. Excellent if you have that team; a liability if you do not. | Fabric |

Fabric is the better fit for: Teams that want Proxmox-style pricing freedom without designing storage, fencing and networking themselves; MSPs and hosters who need a tenant portal and per-tenant networks out of the box; Fleets where one-click, rollback-safe updates across many sites matter; Buyers who need a 24×7 SLA from the vendor.

Proxmox is the better fit for: Home labs, SMBs and teams with strong Linux skills and one or two clusters; Shops that need snapshots and Proxmox Backup Server today; Environments built on ZFS or local storage rather than a distributed pool.

Migration: Proxmox and Fabric both run standard disk-image formats, so VMs move by exporting a disk image and importing it as a Fabric volume. Many teams keep Proxmox for lab and dev and run production tenants on Fabric, where the tenant portal and verified fencing earn their keep.

**Is Proxmox cheaper than Fabric?**

On licence, usually yes: Proxmox VE is free and its Standard subscription for a dual-socket node is about €1,100 per year, versus $1,800 for a Fabric Medium node. Fabric’s free Community edition is also unlimited. The difference you pay for is the integrated appliance — storage policies, per-tenant SDN, verified fencing, atomic fleet updates and a 24×7 option.

**Does Proxmox have multitenancy like Fabric?**

Proxmox has users, pools and permissions on one UI, which works for internal departments. It does not ship a tenant self-service portal with isolated per-tenant networks, DHCP/DNS and quotas; Fabric does, in every class.

**How does HA fencing differ between Proxmox and Fabric?**

Proxmox relies on each node self-fencing through a watchdog when it loses quorum. Fabric additionally powers the failed node off through its BMC via a live peer and blocks it at the storage layer, and refuses to restart a VM elsewhere until that fence is verified.

**Can I run Fabric on the same hardware I run Proxmox on?**

Almost certainly. Fabric targets commodity x86-64 servers with any mix of SAS, SATA or NVMe drives and has no restrictive compatibility list.

**What does Proxmox do that Fabric does not yet?**

VM snapshots, scheduled node-to-node replication and Proxmox Backup Server are available today; Fabric’s snapshots and cross-site replication are roadmap items. Proxmox also has a far larger community and third-party tooling.


### Fabric vs Nutanix (Nutanix Cloud Infrastructure (NCI) with the AHV hypervisor, Nutanix)

URL: https://fabric.izt.cloud/compare/fabric-vs-nutanix

Nutanix proved that HCI could be an appliance experience. It also proved how expensive per-core tiers get on dense servers. Fabric keeps the appliance idea — one image, self-healing, plain-language health — and drops the meter and the feature matrix.

- Nutanix licenses per core in Starter, Pro and Ultimate tiers, with networking security, DR orchestration and other capabilities gated by tier or add-on. Fabric has one edition, priced per node by class.
- Nutanix runs a controller VM on every node that owns storage and consumes host RAM and CPU; Fabric’s storage and agent run in the node OS itself.
- On a 32-core node, reseller-listed NCI Pro is roughly eleven times Fabric’s Medium class per year. Nutanix’s data services and DR remain more mature.

Pricing: Nutanix — Per physical core, subscription, in Starter / Pro / Ultimate tiers with support level multipliers; Prism Central, Flow and DR features vary by tier or add-on. Example: ~$20,800 / node / yr (NCI Pro with Production support at ~$649/core reseller list, 32 cores, 2026). Fabric — Fabric Medium class: $1,800 per node per year, everything included.

| Dimension | Fabric | Nutanix | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Per core, subscription, three tiers plus add-ons; feature availability depends on the tier you buy. | Fabric |
| What you install | One signed firmware image per node plus a control-plane appliance; the storage layer and node agent run in the node OS. | AHV plus a Controller VM (CVM) on every node that owns the storage fabric, and Prism Central as a separate management VM cluster. | Fabric |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | Distributed storage fabric with deduplication, compression, erasure coding and data locality — deep, mature data services. | Nutanix |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | Flow Virtual Networking (overlay) and Flow Network Security (micro-segmentation), available in higher tiers or as add-ons. | Even |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | VM HA with reserved capacity and automatic restart; well-proven. | Even |
| Live migration | Live migration with pre-built tenant networking; hibernate and resume anywhere. | Live migration and Acropolis Dynamic Scheduling for automatic balancing. | Nutanix |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Life Cycle Manager (LCM) orchestrates one-click upgrades of AOS, AHV, firmware and Prism — polished, but a multi-component stack. | Even |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Projects and role-based access in Prism; a self-service portal for tenants historically required Prism Central plus higher tiers. | Fabric |
| Snapshots and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap and will not be tier-gated. | Snapshots, async and near-sync replication, Metro Availability and DR orchestration — a market leader, partly tier-gated. | Nutanix |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Dark-site deployments are supported with a dark-site LCM bundle; licensing is applied via portal-generated files. | Fabric |
| Hardware | Commodity x86-64, any drive mix, no restrictive compatibility list. | Nutanix NX appliances or validated OEM platforms from a compatibility list. | Fabric |
| Price at density | A 64-core node is Large class at $3,200; doubling cores moves one band. | Doubling cores doubles the licence: roughly $41,500 at 64 cores at the same per-core rate. | Fabric |
| Ecosystem and maturity | New platform; REST API, Prometheus metrics; backup-ISV interfaces on the roadmap. | Broad ISV and backup ecosystem, large installed base, Kubernetes and database services layered on top. | Nutanix |

Fabric is the better fit for: Buyers who like the Nutanix appliance experience but not per-core tiers; Mid-market and edge estates where a $20k-per-node licence is out of reach; Sovereign or disconnected sites that cannot depend on a licensing portal; MSPs who want tenancy included rather than tier-gated.

Nutanix is the better fit for: Enterprises that need Metro Availability, near-sync DR or advanced data services today; Shops standardised on Nutanix’s database and Kubernetes services; Sites that want vendor-supplied appliance hardware.

Migration: Nutanix AHV VMs export to standard disk images, which import as Fabric volumes. The usual pattern is a new Fabric cluster on commodity hardware alongside the Nutanix estate, then a tier-by-tier move timed to the Nutanix renewal.

**How does Nutanix pricing compare with Fabric?**

Nutanix charges per core in tiers. Reseller-listed NCI Pro at about $649 per core makes a 32-core node roughly $20,800 per year; the same node is a Fabric Medium at $1,800 with every feature included.

**Does Fabric have a controller VM like Nutanix?**

No. Fabric’s storage layer and node agent run directly in the signed node OS, so no per-node controller VM reserves host CPU and RAM.

**Does Fabric gate features by tier the way Nutanix Starter, Pro and Ultimate do?**

No. Every Fabric class — and the free Community edition — is the whole platform. Classes differ only by the node’s core band.

**What does Nutanix do better than Fabric?**

Data services and disaster recovery: deduplication, erasure coding, near-sync replication and Metro Availability are mature today, while Fabric’s snapshots and cross-site replication are roadmap items. Nutanix also has a far larger ISV ecosystem.


### Fabric vs Scale Computing (Scale Computing SC//Platform (HyperCore), Scale Computing)

URL: https://fabric.izt.cloud/compare/fabric-vs-scale-computing

Scale Computing made edge HCI simple and reliable, and it is the platform we respect most in that segment. It also meters per core, sells a cut-down Standard tier, caps cluster size and ships no tenancy or SDN. Fabric was built for the same edge sites with those four constraints removed.

- Scale licenses per core, channel-only, in Standard and Professional tiers; Standard drops replication, snapshot scheduling, GPU and affinity. Fabric has one edition per node class.
- Scale’s storage is a fixed two-copy mirror with no tiering policy; Fabric storage policies set copies, tiers, failure domain and QoS per disk.
- Scale has snapshots, replication and a local witness today; Fabric’s are roadmap items. Fabric has multitenancy and per-tenant SDN today; Scale has neither.

Pricing: Scale Computing — Per physical core, subscription (1/3/5 years), channel-only; Standard vs Professional tiers, Fleet Manager licensed separately per cluster. Example: ~$6,400 / node / yr (verified 5-year street rate of ~$201/core from reseller catalogs and a transacted public-sector contract, 32 cores, August 2026). Fabric — Fabric Medium class: $1,800 per node per year; Edge class $500 for ≤8-core boxes, the size Scale most often sells.

| Dimension | Fabric | Scale Computing | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Per core, subscription, channel-only. Standard is the cut-down tier; Fleet Manager is a separate per-cluster licence. | Fabric |
| What you install | One signed firmware image per node plus a control-plane appliance you host; nodes have no inbound ports. | HyperCore, a single-image appliance OS with a built-in HTML5 UI — genuinely simple, with management inside the cluster. | Even |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | SCRIBE: a fixed two-copy mirror across the cluster with automatic tiering between flash and disk. No policy per disk, no adjustable replica count or rack domain. | Fabric |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | VLAN tagging on virtual NICs. No overlay SDN, no tenant DHCP/DNS, no micro-segmentation. | Fabric |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | Automatic VM restart on node failure; a local witness for two-node edge sites. | Even |
| Live migration | Live migration; hibernate and resume on any node. | Live migration within the cluster. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Rolling, non-disruptive HyperCore updates — one of its best features. | Even |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | None. One organisation per cluster; role-based users on the same UI. | Fabric |
| Snapshots and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap and will be in every class. | Scheduled snapshots and cross-cluster replication built in — but only in the Professional tier. | Scale Computing |
| Cluster size | One node to many racks in one fabric; rack-aware placement kicks in automatically at three racks. | Eight nodes per cluster; larger estates are many clusters under Fleet Manager. | Fabric |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Clusters run offline; Fleet Manager is a cloud service. | Fabric |
| Hardware | Bring your own commodity x86-64 servers, any drive mix. | Scale-supplied appliances or a short list of certified platforms; hardware is part of the deal. | Fabric |
| Fleet management | (Roadmap) Multi-fabric Global view is planned; today each fabric is managed from its own appliance. | Fleet Manager (cloud) for many clusters — mature, separately licensed. | Scale Computing |
| Guest monitoring | One-second VM telemetry and guest inventory from the guest agent, in the portal and over Prometheus. | AIME autonomous monitoring with guest-level insight. | Even |

Fabric is the better fit for: Retail, branch and industrial edge that wants per-node pricing on its own hardware; Operators who need tenancy or per-tenant networks at the edge; Sites that outgrow an eight-node cluster; Air-gapped estates that cannot use a cloud fleet manager.

Scale Computing is the better fit for: Buyers who want hardware and software from one vendor with one support number; Sites that need snapshots and replication today; Very large multi-cluster fleets managed centrally.

Migration: HyperCore exports VMs as standard disk images; Fabric imports them as volumes. Because Fabric runs on commodity servers, the Scale appliances can stay in place until their term ends while new sites deploy on Fabric.

**How does Scale Computing pricing compare with Fabric?**

Scale licenses per core through resellers; a verified 5-year street rate of about $201 per core makes a 32-core node roughly $6,400 per year. Fabric’s Medium class is $1,800 per node per year, and its Edge class is $500 for boxes of eight cores or fewer.

**Does Scale Computing support multitenancy?**

No. SC//Platform serves one organisation per cluster. Fabric includes tenants, quotas, isolated tenant networks and a tenant self-service portal in every class.

**Is Fabric as simple to run as Scale HyperCore?**

That is the goal. Health is Healthy / Degraded / Critical, disks are disks and control nodes are control nodes; the rolling firmware update is one click. Where Scale ships the hardware too, Fabric expects you to bring commodity servers.

**What does Scale Computing do that Fabric does not yet?**

Scheduled snapshots, cross-cluster replication and a cloud Fleet Manager are available today (the first two in the Professional tier); Fabric’s equivalents are roadmap items.


### Fabric vs VergeIO (VergeOS by VergeIO, VergeIO)

URL: https://fabric.izt.cloud/compare/fabric-vs-vergeio

VergeIO is the platform whose licensing model looks most like ours: one licence per physical server, no core or capacity counting. The difference is the number on the invoice, and how each platform builds the stack underneath.

- Both license per server with all features included. Reseller-listed VergeOS Enterprise runs roughly $12,000 per node per year; Fabric’s Medium class is $1,800 and Edge is $500.
- VergeOS nests virtual datacenters and has a mature snapshot/replication story; Fabric ships tenants and a tenant portal today with snapshots and replication on the roadmap.
- Fabric is delivered as a signed A/B firmware image with automatic rollback and offers a free unlimited Community edition.

Pricing: VergeIO — Per physical server, subscription, all features included; SMB bundles with VM caps and Enterprise per-node terms. Example: ~$12,000 / node / yr (VergeOS Enterprise per-node reseller pricing, 2026 (a 5-year Enterprise term lists at ~$61,760 per node)). Fabric — Fabric Medium class: $1,800 per node per year; Edge $500; Community free.

| Dimension | Fabric | VergeIO | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Per server, all-inclusive subscription. SMB bundles cap VM counts; Enterprise is per node. | Even |
| Price | $500 – $5,000 per node per year by class; free unlimited Community edition. | Roughly $12,000 per node per year at Enterprise reseller pricing; no free edition. | Fabric |
| What you install | One signed, read-only firmware image per node and a control-plane appliance. | VergeOS, a single Linux-based OS per node with management built in. | Even |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | vSAN-style integrated storage with global deduplication and snapshots as a core primitive. | Even |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | Integrated virtual networking with per-VDC routers, firewalls and NAT — strong. | Even |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | Automatic VM restart on node failure. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Rolling OS updates across the cluster. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Nested virtual datacenters (VDCs) — tenants can run tenants — with delegated administration. The most complete tenancy model in the field. | VergeIO |
| Snapshots and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap. | Snapshots at VM, VDC and system level; ioGuardian and site-to-site sync built in. | VergeIO |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Supports disconnected sites. | Even |
| Hardware | Commodity x86-64, any drive mix, no restrictive compatibility list. | Commodity x86, including reuse of existing VMware hosts. | Even |

Fabric is the better fit for: Buyers who want per-server licensing at a fraction of VergeIO’s price; Edge and mid-market estates where $12,000 per node is not viable; Teams that value signed, rollback-safe firmware updates; Anyone who wants to start free on unlimited nodes.

VergeIO is the better fit for: Service providers who need nested tenancy (tenants of tenants); Sites that need snapshot-based DR and site sync today; Larger enterprises comfortable with the price.

Migration: Both platforms import standard disk images; VergeOS estates typically move workload by workload at renewal. Fabric’s Community edition lets you validate the platform on the target hardware before spending anything.

**VergeIO also charges per server — why is Fabric cheaper?**

Fabric’s classes start at $500 per node per year and top out at $5,000 for 65+ core hosts, with a free unlimited Community edition. VergeOS Enterprise lists at roughly $12,000 per node per year through resellers.

**Does Fabric have nested virtual datacenters like VergeOS?**

No. Fabric has flat tenants with quotas, isolated networks and a self-service portal — enough for an MSP or an internal cloud, but not tenants-of-tenants.

**Which is more mature on disaster recovery?**

VergeOS: snapshots, ioGuardian and site sync are shipping. Fabric’s snapshots and cross-site replication are roadmap items and will be included in every class.


### Fabric vs Harvester (Harvester HCI (SUSE Virtualization), SUSE)

URL: https://fabric.izt.cloud/compare/fabric-vs-harvester

Harvester runs virtual machines as Kubernetes objects on top of a container orchestrator. That is exactly right if your world is Kubernetes. Fabric is built the other way round: a VM-first appliance that can hand storage to Kubernetes later, with an operator surface that never asks you to learn the internals.

- Harvester is free and open source, supported through SUSE subscriptions; Fabric is free in Community and per node by class when supported.
- Harvester’s storage, networking and HA are Kubernetes primitives; Fabric’s are appliance features — storage policies, per-tenant SDN, verified fencing.
- Harvester wins for platform teams standardised on Rancher; Fabric wins for infrastructure teams who need a tenant portal, a 24×7 SLA and fleet-wide atomic updates.

Pricing: Harvester — Free and open source (Apache 2.0). Enterprise support through SUSE subscriptions, typically alongside Rancher Prime. Example: Free software; support priced by SUSE subscription (public SUSE pricing is quote-based, 2026). Fabric — Fabric Community is free and unlimited; Medium class $1,800 per node per year with business-hours support, 24×7 at +30%.

| Dimension | Fabric | Harvester | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Open source, free; enterprise support via SUSE subscription, quote-based. | Even |
| What you install | A signed firmware image per node plus a control-plane appliance; the operator never sees a container orchestrator. | An immutable OS image running a Kubernetes distribution, with VMs as custom resources; operated through Rancher. | Even |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | Replicated block volumes managed by the cluster; per-volume replica count, no tiering policy or failure-domain model across racks. | Fabric |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | VLAN networks and cluster networking via the container network layer; tenant DHCP, gateways and micro-segmentation are Kubernetes policy work. | Fabric |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | VM rescheduling when a node is declared not ready, relying on volume replicas; no out-of-band power fencing. | Fabric |
| Live migration | Live migration; hibernate and resume on any node. | Live migration supported. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Cluster-wide upgrades of the OS and orchestrator that must be planned carefully on small clusters. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Namespaces and Rancher projects with RBAC — good for platform teams, not a tenant self-service portal. | Fabric |
| Snapshots and backup | (Roadmap) Snapshots and cross-site replication are on the roadmap. | Volume snapshots and VM backups to S3/NFS built in. | Harvester |
| Kubernetes integration | (Roadmap) A CSI driver exposing storage policies to Kubernetes is on the roadmap; VMs and containers side by side. | Native: Harvester is Kubernetes, with guest clusters provisioned through Rancher. | Harvester |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Air-gapped installs are supported with a private registry. | Even |
| Hardware | Commodity x86-64, any drive mix. | Commodity x86-64; heavier baseline resource footprint per node. | Fabric |

Fabric is the better fit for: Infrastructure teams who run VMs first and want an appliance, not an orchestrator; MSPs and internal clouds that need a tenant portal; Sites that need verified fencing and rollback-safe fleet updates; Buyers who want a vendor 24×7 SLA per node.

Harvester is the better fit for: Platform teams standardised on Rancher and Kubernetes; Shops that want VMs and containers under one Kubernetes API today; Open-source-first organisations with SUSE relationships.

Migration: Harvester VMs export as disk images and import into Fabric as volumes. Teams that keep Kubernetes as the application layer can later consume Fabric storage policies through the planned CSI driver.

**Is Fabric built on Kubernetes like Harvester?**

No. Fabric is a VM-first appliance: a signed node image, a control-plane appliance and a REST API. A CSI driver to serve Fabric storage to Kubernetes clusters is on the roadmap.

**Which is easier to operate for a traditional virtualization team?**

Fabric. Its operator surface speaks in tenants, VMs, disks, policies and Healthy / Degraded / Critical health. Harvester expects familiarity with Kubernetes concepts and Rancher.

**Is Harvester cheaper?**

Both have free editions. Harvester support comes through SUSE subscriptions; Fabric support is per node by class from $500 per year with everything included.


### Fabric vs Azure Local (Microsoft Hyper-V and Azure Local (formerly Azure Stack HCI), Microsoft)

URL: https://fabric.izt.cloud/compare/fabric-vs-azure-local

Azure Local is Microsoft’s HCI, billed per core per month through an Azure subscription and designed to stay connected to it. Hyper-V on Windows Server is the on-prem classic. Fabric is for the sites that must run whether or not the cloud does.

- Azure Local bills per physical core per month via Azure and must sync with Azure regularly; Fabric is per node per year and needs no connection at all.
- Azure Local runs only on validated hardware from the Azure Local catalog; Fabric runs on commodity servers.
- Microsoft wins for Windows-centric estates and Azure hybrid services; Fabric wins for sovereignty, mixed Linux/Windows fleets and edge sites without an Azure relationship.

Pricing: Azure Local — Azure Local: per physical core per month, billed through an Azure subscription, plus Windows Server guest licensing. Hyper-V: included with Windows Server Datacenter, per core with a 16-core minimum per server. Example: Per-core monthly Azure billing plus guest licences (Microsoft publishes regional per-core rates; totals depend on Azure agreement and guest OS licensing, 2026). Fabric — Fabric Medium class: $1,800 per node per year including the whole platform; guest OS licensing is unchanged either way.

| Dimension | Fabric | Azure Local | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Azure Local: per core per month through Azure. Windows Server Datacenter: per core, 16-core minimum, for unlimited Hyper-V guests. | Fabric |
| Cloud dependency | None. No registration, no sync, no kill-switch. | Azure Local must register with Azure and sync at least every 30 days to stay in a supported, billable state. | Fabric |
| What you install | One signed firmware image per node plus a control-plane appliance. | Azure Local OS on validated nodes, managed through Azure Arc and the Azure portal; or Windows Server with Hyper-V, Failover Clustering and System Center / Windows Admin Center. | Fabric |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | Storage Spaces Direct: mirror or parity resiliency, tiering, deduplication — solid, and tied to validated hardware. | Even |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | Network ATC and optional SDN (Network Controller, gateways) — capable but complex to deploy. | Fabric |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | Failover Clustering with quorum witnesses — mature and well understood. | Even |
| Live migration | Live migration; hibernate and resume on any node. | Hyper-V live migration and storage migration. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Cluster-Aware Updating; Azure Local lifecycle via Azure Update Manager. Monthly Windows patching cadence. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Azure RBAC over Arc-enabled resources; no tenant self-service portal on-prem. | Fabric |
| Snapshots and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap. | Checkpoints, Hyper-V Replica and Azure Site Recovery integration. | Azure Local |
| Hardware | Commodity x86-64, any drive mix, no restrictive compatibility list. | Azure Local requires validated systems from the catalog; Hyper-V is broader. | Fabric |
| Windows guest economics | Guest licensing is your existing Microsoft agreement; no advantage or penalty. | Datacenter edition covers unlimited Windows guests; Azure Hybrid Benefit and AVMA can be compelling for Windows-only estates. | Azure Local |
| Ecosystem | New platform; REST API, Prometheus metrics. | Deep Microsoft tooling, Azure hybrid services (Arc, Backup, Site Recovery, Monitor). | Azure Local |

Fabric is the better fit for: Air-gapped, sovereign or regulated sites that cannot register with a public cloud; Mixed Linux/Windows estates on commodity hardware; Edge sites with no Azure agreement or unreliable connectivity; Buyers who want one predictable annual bill per node.

Azure Local is the better fit for: Windows-only estates that already own Datacenter licences; Organisations committed to Azure hybrid services; Teams standardised on Microsoft tooling.

Migration: Hyper-V VHDX disks convert to standard images and import into Fabric as volumes; Windows guests keep their existing licences. Sites often start with Fabric at the edge where Azure connectivity is weakest.

**Can Azure Local run disconnected like Fabric?**

Azure Local is designed to stay registered with Azure and must sync periodically; extended disconnection is a special case. Fabric never needs a connection.

**Does Fabric run on the same servers as Azure Local?**

Yes — and on many that Azure Local does not support, since Fabric has no validated-hardware requirement.

**Is Hyper-V free?**

Hyper-V is a role of Windows Server, licensed per core with a 16-core minimum per server; the standalone free Hyper-V Server was discontinued after the 2019 release. Fabric’s Community edition is free and unlimited.


### Fabric vs XCP-ng (XCP-ng with Xen Orchestra (Vates), Vates)

URL: https://fabric.izt.cloud/compare/fabric-vs-xcp-ng

XCP-ng is a mature open-source hypervisor with an excellent management and backup layer in Xen Orchestra. What it is not, out of the box, is a hyperconverged platform: shared storage is an add-on and multitenancy is not the model. Fabric starts from the other end.

- XCP-ng is free; Vates sells per-host support and Xen Orchestra Appliance plans. Fabric is free in Community and per node by class when supported.
- Xen Orchestra’s built-in backup, continuous replication and VMware import are shipping strengths; Fabric’s equivalents are roadmap items.
- Fabric ships distributed storage, per-tenant SDN, verified fencing and a tenant portal as the product; XCP-ng pools top out at 16 hosts and add hyperconvergence via a separate storage layer.

Pricing: XCP-ng — Free hypervisor (GPL). Vates support per host and Xen Orchestra Appliance subscription plans; hyperconverged storage (XOSTOR) supported as an add-on. Example: Free software; support per host via Vates (Vates publishes tiered plans; totals depend on host count and plan, 2026). Fabric — Fabric Community is free and unlimited; Medium class $1,800 per node per year with everything included.

| Dimension | Fabric | XCP-ng | Edge |
|---|---|---|---|
| Licensing model | Per physical node by node class (five core bands: ≤8 / 9–16 / 17–32 / 33–64 / 65+), every feature in every class, free unlimited Community edition. No per-core meter, no feature matrix. | Free hypervisor; per-host support and orchestrator subscriptions from Vates. | Even |
| What you install | One signed firmware image per node plus a control-plane appliance. | XCP-ng on each host plus a Xen Orchestra virtual appliance for management and backup. | Even |
| Storage | Distributed block storage pooled across every node, driven by storage policies: ordered device tiers, copies and minimum copies, host- or rack-level failure domains, per-disk IOPS/bandwidth QoS and inline compression with in-place re-pack. No SAN and no storage licence. | Shared storage over NFS/iSCSI/FC or local; hyperconverged storage via XOSTOR, a separately supported layer. | Fabric |
| Networking | Per-tenant SDN in the platform: overlay or VLAN-backed virtual networks, dual-stack IPv4/IPv6, built-in DHCP/DNS, micro-segmentation policies, distributed stateless gateways and ARP or BGP announcement of public prefixes. Not a separate SDN licence. | VLANs and Open vSwitch bridges; SDN controller for private networks across hosts; no tenant DHCP/DNS or gateways. | Fabric |
| High availability | Acknowledged 10-second heartbeats, a liveness verdict in about 20 seconds, and no restart elsewhere until the failed node is verifiably fenced — powered off through its own BMC and blocked at the storage layer. | Pool HA with a heartbeat storage repository and host self-fencing. | Fabric |
| Live migration | Live migration; hibernate and resume on any node. | Live migration within and between pools, including storage migration. | Even |
| Updates and lifecycle | One signed A/B firmware image per node carries the hypervisor, storage, networking and agent. A rolling update evacuates each node by live migration, flips its image slot and rolls back automatically if it does not come up healthy. | Rolling pool updates orchestrated from Xen Orchestra; host-level packages. | Fabric |
| Multitenancy | Built in. Tenants, quotas, isolated per-tenant networks with their own DHCP/DNS, and a tenant self-service portal with browser consoles — the same product an MSP bills monthly per node. | Users, groups and ACLs with a self-service section in Xen Orchestra; no isolated per-tenant networks by default. | Fabric |
| Backup and DR | (Roadmap) Snapshots and cross-site replication are on the roadmap. | Xen Orchestra backup, delta backup, continuous replication and disaster-recovery jobs built in — a standout. | XCP-ng |
| Pool and cluster size | One node to many racks in one fabric. | Sixteen hosts per pool; larger estates are multiple pools. | Fabric |
| Air-gapped operation | Designed for it: local sign-in (password, TOTP, passkeys — never an emailed link), a control plane on your own appliance, signed updates imported and verified on site, no phone-home, no kill-switch. | Runs offline; update repositories can be mirrored. | Even |
| VMware migration | (Roadmap) Turnkey ESXi import is on the roadmap. | Xen Orchestra’s V2V import from VMware is shipping and widely used. | XCP-ng |
| Hardware | Commodity x86-64, any drive mix. | Broad x86 support; hyperconverged storage needs matched local disks. | Even |

Fabric is the better fit for: Teams that want hyperconverged storage and networking as the product, not an add-on; MSPs who need a tenant portal and per-tenant networks; Fleets beyond sixteen hosts per cluster; Buyers who want verified fencing and rollback-safe firmware updates.

XCP-ng is the better fit for: Shops with existing shared storage arrays; Teams that need built-in backup and VMware import today; Former XenServer / Citrix Hypervisor estates.

Migration: XCP-ng exports VMs in standard formats that import into Fabric as volumes. Sites with a SAN often keep XCP-ng for those workloads and place new, storage-heavy or tenant-facing workloads on Fabric.

**Is XCP-ng hyperconverged?**

Not by default. XCP-ng is a hypervisor; hyperconverged storage comes from XOSTOR, a separately supported layer, or from external shared storage. Fabric pools every node’s disks into policy-driven distributed storage as the base product.

**Which has better backup?**

Xen Orchestra: delta backup, continuous replication and DR jobs are shipping today. Fabric’s snapshots and cross-site replication are roadmap items.

**How big can a cluster get?**

XCP-ng pools are limited to sixteen hosts. A Fabric fabric grows from one node to many racks, with rack-aware placement engaged automatically at three racks.


## Contact

Website: https://fabric.izt.cloud/ · Company: https://izt.cloud · Pricing and free Community edition: https://fabric.izt.cloud/pricing