VergeOS Architecture: One Code Base for the Entire Data Center
VergeOS is a private cloud OS. Compute, storage, networking and data protection run as one integrated platform, on shared servers, built by one development team. A hypervisor swap changes the logo on the renewal. VergeOS removes the five-product stack underneath it.
The Wrong VMware Exit Costs More Than VMware Did
A like-for-like hypervisor swap moves the renewal pain to a different vendor and leaves the operational architecture intact. The hypervisor changes. The five-product stack underneath it stays.
Reconciling that stack every time one product changes consumes 40% to 60% of an infrastructure team’s working hours, by industry telemetry. The exit worth doing consolidates the stack itself. It does not re-license it.
86%
of organizations are actively reducing their VMware footprint.
4%
have fully migrated off VMware.
41%
report CEO or CFO pressure to make the move.
74%
are evaluating alternatives, most of them hypervisor swaps.
State of the VMware Exit, 2026
Five Products or One Code Base
A private cloud OS is not a hypervisor. It is the operating system for the data center. Compute, storage, networking and data protection ship, upgrade and receive support as one product.
Five products, integrated by you
- Developed independently
- Licensed separately
- Supported individually
- Every upgrade is a coordination exercise
One platform, integrated by design
- One code base, one OS
- One license
- One support team
- One upgrade path
The Difference Is Not Features. It Is Scale.
A private cloud OS is structurally simpler than the stack it replaces. It has fewer interfaces, fewer integration boundaries and fewer places for the system to fail.
The conventional stack accumulates complexity at every boundary: hypervisor to HCI, HCI to storage, storage to backup. VergeOS removes the boundaries. The same code that runs compute also runs the file system, the snapshot engine and the tenant isolation layer.
Consolidation Is Measured
These numbers come from VergeOS customer production environments that moved from a VMware-based three-tier stack to VergeOS on the same or commodity hardware.
What Consolidation Looks Like in the Code
VergeOS is a software architecture with measurable consequences in three places: the file system, the code base and the resilience model. A coordinated stack of separate products cannot deliver any of the three.
VergeFS Runs as a Service
VergeFS runs inside VergeHV, the VergeOS hypervisor, not as a storage VM. No extra hop, no separate communication path. The file system shares memory and scheduling with compute, and server-class flash inside the nodes replaces dedicated arrays.
Every Feature Shares State
Deduplication works across compute, storage, networking and replication. Snapshots are an OS primitive for a VM, a Virtual Data Center or a tenant. High availability is the platform’s default behavior, with no separate controller.
Replication Plus Inline Repair
Distributed synchronous replication handles drive and server failures. ioGuardian adds an inline repair server that feeds missing data to running VMs in real time when failures exceed the protected state.
The Data Center as a Portable Object
A virtual machine packages one workload. A Virtual Data Center (VDC) packages an entire data center: VMs, storage, network configuration, IP addressing, security policies, user permissions and operational settings. VergeOS creates, clones, moves, snapshots, replicates and recovers that environment as one object.
- Multi-tenancy. Service providers give each customer its own VDC with no visibility into any other.
- Application isolation. Enterprises separate production, development and departmental workloads on shared infrastructure.
- Blast radius containment. A failure or attack inside one VDC stays inside that VDC.
- Environment-level recovery. Restore the whole known-good environment, not individual parts.
- Disaster recovery. Replicate a VDC to a second site with its full operational context.
One VDC replicates with its networking, IP addressing and policies intact.
Protection Starts at the Write
When a VM writes to VergeFS, global deduplication checks the block first. A duplicate updates metadata only. A new block is written synchronously to multiple nodes before VergeOS acknowledges the write.
- RF2 or RF3. Survive one or two drive or node failures with VMs online and protection rebuilt in the background.
- ioGuardian. One extra server, older hardware and HDDs welcome, holds another copy and serves missing blocks inline. Customers license the server, not the feature.
- IOclone snapshots. Read-only by default, with immutable snapshots that cannot change or be deleted before retention expires. Recover an instance, a VDC, a VM or individual files.
- ioReplicate. Snapshot-driven, WAN-aware replication with global deduplication across the link, so only unique blocks cross.
Multi-layer retrieval keeps data flowing through complex failures.
Recovery Is a Platform Property, Not a Backup Product
The average ransomware dwell time is six days. Recovery posture has to live inside the platform before the encryption starts.
VDC Isolation
Each VDC is an isolated boundary. A compromised tenant does not reach the platform or other tenants, by architecture rather than policy.
IOclone Every 10–15 Minutes
Dedup-aware snapshots at the file system layer. Unchanged blocks cost almost nothing to keep, so RPO is minutes.
ioFortify Detection
Encryption breaks the deduplication curve. ioFortify watches the dedup ratio and alerts within 10 to 15 minutes of an encryption event.
6-Day Window, Covered
With 10-minute snapshots kept for weeks, roll the entire VDC back to any clean interval at the file system layer.
| Recovery Capability | Conventional Stack | VergeOS |
|---|---|---|
| Tenant-level air-gap isolation | Network policy | ✓ Built into the VDC |
| Snapshot frequency without a performance penalty | 4+ hours | ✓ 10–15 minutes |
| Snapshot retention without a performance penalty | Days | ✓ Months |
| Inline encryption-anomaly detection | ✕ | ✓ ioFortify |
| Instant rollback at the file system layer | ✕ | ✓ |
| Single-product upgrade path for the recovery stack | ✕ | ✓ |
| Recovery target included with the platform | ✕ | ✓ |
VergeOS lightens the load on backup. The backup product’s role shifts from primary recovery engine to long-term retention and archive. VergeOS works with Veeam and other UHAPI or oVirt-compatible backup software.
Veeam Stays. VergeOS Adds a Layer Beneath It.
Veeam Backup & Replication 13.1 added VergeOS as a natively supported platform on July 29, 2026, through the Universal Hypervisor API. The jobs, retention policies and console the team already runs stay in place.
Infrastructure Layer · VergeOS
Synchronous replication, IOclone snapshots every 10 to 15 minutes, ioGuardian inline recovery and ioReplicate to a second site. Answers the 10-minute-old mistake, the multi-drive failure and the site outage, with the VM still running.
Workload Layer · Veeam
Long-term retention, independent immutable and offsite copies, application-item recovery and a catalog that outlives the platform. Answers the audit, the seven-year rule and the recovery onto different hardware.
Selectable in the Console
Add a VergeOS Manager under Virtualization Platforms the same way you add a vCenter. Changed Block Tracking, Backup Scan and application-aware backup are included.
Restore in Both Directions
Entire VM Restore moves workloads into VergeOS from backups of vSphere, Hyper-V, Nutanix AHV, Proxmox VE, public cloud or physical servers, and back out again if the evaluation says so.
Setup takes one step on each side: turn on the oVirt API compatibility package on VergeOS 26.1.7 or later, then add the VergeOS Manager in Veeam 13.1.
Rancher Stays. VergeOS Replaces the Substrate.
VergeOS does not ship a competing Kubernetes distribution. Four Helm charts on the verge-io GitHub repository delegate directly to the VergeOS API, and the management plane the team runs today keeps running. RKE2, K3s, upstream Kubernetes and vendor distributions all run on VergeOS.
- CSI driver. Persistent volumes live on VergeFS with inline deduplication, multi-tier placement and integrated snapshots. RWO through block, RWX through EXT4-over-NFS.
- Cloud Controller Manager. LoadBalancer services map to VergeOS VNet NAT rules, so MetalLB is no longer required.
- Cluster Autoscaler. Node pools track pending pod requests, and VergeOS adds or removes the VMs.
- Rancher node driver and UI extension. The Rancher flow on VergeOS matches the flow on vSphere.
The conventional VMware-plus-Kubernetes stack licenses the same workload three times: hypervisor, distribution and overlay storage. VergeOS collapses the three into one platform, and server count drops 30% to 40% for the same density.
Four Helm charts on the verge-io GitHub repository.
Sizing, Licensing and Testing Change Shape
Size the Workload Once
Plan on 30% to 40% fewer servers. The array goes away, the storage VM overhead goes away and one shared cache replaces two competing ones. A four-server VergeOS cluster typically replaces a six- or seven-server VMware cluster.
Per Server, Everything Included
Licensed per physical server, not per socket, core, terabyte or IOPS. Replication, snapshots, ioGuardian, ioFortify, ioMigrate, ioReplicate and VergeIQ ship in the same license. Dense hardware costs no more to license.
DR Tests in Under an Hour
DR recovery is a three-click VDC operation, so customers test as a routine and report a 100% DR test success rate. Upgrades roll one node at a time with zero downtime, from one code base on one cadence.
What You Remove, What Stays and What Gets Absorbed
VergeOS is one integrated platform that ends the need for the products below. Primary replacements come first, supporting roles second.
| Product Category | Role | What VergeOS Delivers |
|---|---|---|
| VMware ESXi / vSphere | Primary | VergeHV, the integrated hypervisor, with live migration, HA, snapshots and cluster scheduling. No separate licensing layer. |
| Nutanix AOS / vSAN / HCI software | Primary | VergeFS global file system with deduplication, compression and erasure coding. |
| External arrays (Pure, NetApp, Dell) | Primary | Tier-1 NVMe and capacity tiers inside the same cluster. Existing Fibre Channel LUNs join VergeFS as a tier until the array retires. |
| VMware NSX / Cisco ACI | Primary | Per-tenant VDCs with isolated networks, microsegmentation and quota-driven resource control. |
| Rubrik / Cohesity backup | Primary | Continuous IOclone snapshots with native replication to a VergeOS DR target. No agents on guest VMs. |
| Veeam Backup & Replication | Supports | Natively supported in VBR 13.1. IOclone and ioGuardian add infrastructure-level recovery beneath the Veeam jobs already in place. |
| VMware Site Recovery Manager | Supports | Built-in DR orchestration. Declare a recovery target, replicate the VDC and fail over. |
| VMware Aria Operations / vRealize | Supports | ioCommand reports health, performance and capacity across compute, storage and the file system in one view. |
| Cloud bursting / DR-as-a-Service | Supports | Replicate to peer VergeOS environments on-premises, in colocation or at a service provider. |
60 to 120 Days on Existing Hardware
A private cloud OS migration needs no forklift, no new servers, no new array and no six months of professional services. The median time from kickoff to full cutover is 60 to 120 days, and teams reach productivity in one to two weeks. The rest is workload pacing.
ioMigrate rehosts VMware VMs in batches, networking included. The team validates each batch in production and retires the legacy environment as workloads move. A Veeam Entire VM Restore offers a second path from any platform Veeam protects.
Certified on Existing Hardware
- Mix server generations, CPU vendors and drive layouts in one instance. VergeOS has never deprecated hardware.
- Combine HCI, storage-only and compute-only nodes to match the hardware you own.
- VxRail customers repurpose existing nodes.
- Minimum: 64-bit Intel or AMD, one 320GB NVMe for metadata, one flash drive for VM storage and 8GB of RAM plus 1GB per TB.
VergeOS Capabilities
Pair Two Proofs of Concept. Compare the Operational Math.
Run VergeOS on the same hardware you would give a hypervisor-swap candidate. Over 60 days, measure three numbers: servers required for the same workloads, recovery time after a tenant-level failure and team hours spent on platform coordination. The answer is arithmetic. POC clusters install in under four hours.
Read the full architecture: Beyond the Hypervisor Swap white paper · Lightboard session
Request a Test Drive
Get hands-on access to a live VergeOS environment.
Frequently Asked Questions About the VergeOS Architecture
What is the VergeOS architecture?
VergeOS is a private cloud OS. Compute, storage, networking and data protection run from one code base of under 400,000 lines, built by one development team. It ships, upgrades and receives support as one product instead of five.
How is a private cloud OS different from a hypervisor swap?
A hypervisor swap replaces the virtualization layer and keeps the rest of the stack. A private cloud OS consolidates virtualization, storage, networking, data protection and disaster recovery into a single code base. The hypervisor swap addresses licensing. The private cloud OS removes the coordination work across separate products.
How does VergeFS differ from storage that runs in a VM?
VergeFS runs as a service inside VergeHV, the VergeOS hypervisor, not as a storage VM. There is no extra hop and no separate communication path. VergeOS delivers more than one million IOPS at 64K per cluster with sub-millisecond latency on commodity NVMe.
What is a Virtual Data Center?
A Virtual Data Center (VDC) packages VMs, storage, network configuration, IP addressing, security policies and permissions into one object. VergeOS clones, snapshots, replicates and recovers that environment as a unit, and an attack inside one VDC stays inside it.
How does VergeOS protect against ransomware?
VDC isolation contains the blast radius. IOclone snapshots run every 10 to 15 minutes and are read-only by default, with an immutable option. ioFortify watches the deduplication ratio and alerts within 10 to 15 minutes of an encryption event, and the VDC rolls back to any clean snapshot.
Does VergeOS work with Veeam?
Yes. Veeam Backup & Replication 13.1, released July 29, 2026, supports VergeOS natively through the Universal Hypervisor API. VergeOS protects the infrastructure, Veeam protects the workload, and Entire VM Restore runs in both directions.
Do I need new hardware to run VergeOS?
No. VergeOS runs on existing Dell VxRail, HPE ProLiant, Cisco UCS, Lenovo ThinkSystem and commodity x86 servers. Mixed generations and node types share one instance, and customers typically run 30% to 40% fewer servers after consolidation.
How long does a migration to VergeOS take?
The median time from kickoff to full cutover is 60 to 120 days. ioMigrate rehosts VMware VMs in batches, and teams familiar with VMware reach productivity on VergeOS in one to two weeks.
How is VergeOS licensed?
VergeOS is licensed per physical server, not per socket, core, capacity or IOPS. Replication, snapshots, ioGuardian, ioFortify, ioMigrate, ioReplicate and VergeIQ ship in the same license.
Does VergeOS replace my Kubernetes distribution?
No. Four Helm charts on the verge-io GitHub repository provide a CSI driver, a Cloud Controller Manager, a Cluster Autoscaler and a Rancher node driver. Rancher stays as the management plane, and VergeOS replaces the vSphere substrate underneath it.
Consolidate the Stack, Not Just the Hypervisor
One code base for compute, storage, networking and data protection, on the servers you already own.