# NodeWeaver — Complete Site Synthesis

> Machine-readable digest of nodeweaver.io for AI agents and LLMs. Last updated: June 2026.
> Shorter navigation index: https://nodeweaver.io/llms.txt

## What NodeWeaver is

NodeWeaver is the most complete edge computing platform on the market. It runs bare-metal "nanoclouds" anywhere — from a single device on a wind turbine to thousands of nodes across a retail fleet — with zero-touch deployment, autonomous management, full high availability, and integrated SASE on your choice of standard x86 hardware.

The core idea: NodeWeaver takes physically disparate nodes and "weaves" them together into a single, highly available compute cluster. All resources from each server are combined into one logical pool, from which the customer simply requests virtual resources — like a hyperscale cloud platform, but on a smaller scale ("nanoclouds").

Tagline: Computing at the Edge of Anywhere.

## Key differentiators

- **Full HA with only 2 nodes.** Competitors require 3. Both nodes are active and run workloads; on failure, VMs from the failed node automatically fail over to the survivor. Significant cost advantage at 1000s of sites.
- **Perpetual licensing.** One of the only companies still selling perpetual software licenses. One physical server = one license, regardless of cores/RAM/workloads. All features included — no tiers, no per-feature/per-VM/per-core surcharges.
- **Tiny footprint.** Runs in as little as 1 core / 1 GB RAM. Bare-metal performance, no nested virtualization tax.
- **Zero-touch deployment.** Under 10 minutes from power-on to operative nanocloud, no trained operator onsite required (a person only unboxes, plugs cables, inserts USB). Recovers from network-down; repeats the process automatically after power loss.
- **Hardware freedom.** Any standard x86 hardware — rugged edge boxes, micro-clusters, rackmount servers — heterogeneous mixes managed as one fabric.
- **Autonomous operations.** Self-healing, self-balancing, self-optimizing. Designed so sites operate indefinitely with no connection to the outside world (airgap and intermittent uplinks are default conditions, not failure modes).

## The edge challenge (why a datacenter platform doesn't work)

Hard-to-reach locations (no monitor/keyboard/operator) · hundreds or thousands of sites · physical constraints (no racks, no AC, no UPS) · bandwidth limits and disconnection · latency requirements (5G RAN, video analytics, industrial control) · physical security exposure · high availability requirements · data and privacy regulations.

## Architecture (3 layers)

1. **At each site:** NodeWeaver cluster on your hardware — compute, storage, networking in one integrated software layer. No external SAN required. No separate hypervisor or bolt-on orchestrator.
2. **Autonomous operations:** continuous monitoring of hardware and workloads, automatic rebalancing and failure recovery. Soft real-time scheduling guarantees resources for latency-critical applications even under full load.
3. **Fleet-wide control:** Zero-Touch Deployment and EdgeInsight for massive scale. Patented DNSOps and EdgeInsight provide policy-driven rollouts, health visibility, and lifecycle management across the entire fleet. The control plane resides on every cluster (sites fully independent), reachable centrally through an encrypted, DDoS-protected tunnel.

## Platform capabilities

### Virtualization
Everything runs in the legacy hypervisor or thinVMs. Automated placement extracts maximum utilization out of heterogeneous hardware while continuous tuning adapts OS and hypervisor parameters to actual workload behavior in real time. Manual pinning, NUMA topology, affinity/anti-affinity, PCI and network passthrough, live hardware replacement with no downtime. Runs VMs, Kubernetes, and pre-configured appliances side by side.

### Soft real-time
Bounded latency and jitter via progressive attributes: REALTIME, CPU_PARTITION, pinned tasks, custom NUMA topology, hypervisor attributes, passthrough. With REALTIME + CPU_PARTITION alone, 99.99% of samples stay under 150 μs. Under a concurrent stresstest with 96 saturating threads, only twelve samples in one billion exceeded 150 μs. Published benchmark (Lenovo P360 Tiny, i7-12700T, Ubuntu 24.04 VM, 4 vCPU, cyclictest): max latency under CPU stress drops from 3,380 μs (baseline) to 55 μs (+guest RT kernel).

### Software-Defined Storage
MLPerf storage benchmark (single node, Unet3D): 99.55% accelerator utilization at 1 GPU, 98.9% at 10 GPUs, up to 2,332 MB/s I/O throughput — storage that keeps AI accelerators fed.

Built-in distributed filesystem, transparent to workloads. Distributed chunkserver engine with per-tenant encryption at rest. Native snapshots, backup, replication. Survives disk loss, node loss, full power events. Start with a 2-node HA pair and transparently add capacity one node at a time (2 → N).

### Software-Defined Networking & SASE
SDN on every node with integrated Secure Access Service Edge. The control plane resides on every cluster and can be reached through an encrypted, DDoS-protected tunnel — even with flaky or asymmetric uplinks. Distributed firewall, DDoS protection. Wire-line speed for Ethernet up to 100 Gbit/s. Works in limited-connectivity and airgapped environments.

### Security
Designed in alignment with IEC 62443. Continuous (Day-N) security: encrypted tamper-proof firmware, hardened OS, encryption-at-rest, tamper detection and remote wipe, strong workload insulation, encrypted management plane, updates with no reboot (even when airgapped), full logging and traceable/auditable APIs.

### Orchestration & automation
One declarative API for 1 → N clusters. GitOps-native, self-heal, update windows, policy. Supports Terraform, Ansible, and open REST APIs. Checkmk export and rsyslog-format log streaming integrate with Splunk, Datadog, Grafana, or any SIEM. Customer self-service onboarding. Integrates with Azure Arc and AWS SSM.

## Specifications

- Minimum node: 1 core · 1 GB RAM (uCPE: 4 cores / 8 GB recommended with HA)
- Cluster size: 1 → N nodes; no theoretical ceiling
- Realtime: P99.99 < 150 μs under saturation; ~50 μs idle
- Throughput: wire-line speed for Ethernet up to 100 Gbit/s
- Deploy: < 10 minutes power-on to operative, no technical expertise required
- Updates: incremental, no reboot, airgap-capable
- Hardware: any standard x86, any vendor
- Fleet management traffic: under 1 MB/hour with centralized management and logs enabled — or zero, fully airgapped
- Version support: previous major version supported for 2 years after a new major release
- Legacy Windows workloads run in VMs alongside modern containers
- Soft real-time scope: supervisory, inspection, and HMI workloads

## Pricing (list)

**Perpetual (primary model):**
- NodeWeaver License: **$1,500 per node, one-time.** Per physical server regardless of configuration. Every feature included. License never expires; nodes keep running even if support lapses (no kill switch).
- Basic Support: **$300/node/year** — business-hours response, support portal, all updates & patches.
- Enterprise Support: **$600/node/year** — adds 24/7 incident response, named engineer, SLA-backed response times, quarterly health review, priority bug-fix routing.
- Support is optional and renewable annually.

**Subscription (100-node minimum):**
- NodeWeaver + Basic: **$99/node/month**
- NodeWeaver + Enterprise: **$149/node/month**
- Perpetual and subscription can be mixed in one fleet.

**Evaluation:** 30-day evaluation, no node limits, full platform, direct access to a NodeWeaver engineer.

**Airgapped license activation:** a software token is generated on the host and activated locally through the web UI as a host attribute — send the host token, receive the license. No internet connection required.

## Solutions

- **Retail & QSR:** POS, payments, digital signage, video analytics (loss prevention), Wi-Fi/SD-WAN, back-office — consolidated on one 2-node cluster per store. Sales continue during WAN outages; sync resumes when the link returns. Stat highlights: 5–15 min site bring-up, 0 on-site IT, 99.99% local availability, any workload mix. Complies with the platform requirements for PCI DSS using built-in security capabilities: isolated virtual networks + workload insulation (segmentation), distributed firewall (default-deny per workload), per-tenant encryption-at-rest + encrypted traffic (cardholder data protection), tamper-resistant audit logs (rule changes and access attempts), encrypted tamper-proof firmware + hardened OS (platform integrity), demonstrable segmentation for annual assessment.
- **Industrial & Manufacturing:** SCADA, MES, OPC UA, vision inspection (GPU passthrough, deterministic frame budgets), robotics, HMIs — replacing per-machine industrial PC sprawl. Soft real-time for PLCs and control loops. Pre-validated rugged hardware from OnLogic, SuperMicro, Lanner, Lenovo.
- **Managed Service Providers (MSPs):** service delivery platform to deploy, manage, and operate dedicated compute for many customers. Multi-tenant with hard isolation, white-label console, drop-ship onboarding, per-tenant policy and usage export for billing.
- **Edge AI:** inference next to cameras/sensors/machines. Direct PCI passthrough of any PCI device — NVIDIA or Intel GPUs, TPUs, FPGAs, capture cards — to the VM that needs it, at full performance. (Full-device passthrough; fractional GPU/MIG partitioning is not offered.) Fleet-scale model deployment with canary and rollback. Data never leaves the site unless you send it.

## Customers & proof points

- **Carnival Corporation** (Intel case study): modernizing shipboard connectivity services; 14K+ IP endpoints in a single 2-node NFV cluster with full HA.
- **Street Edge** (Intel case study): next-gen edge compute and networking services.
- Customer quotes from a global maritime operator (SATCOM senior architect), a retail chain VP of Technology, an EU packaging manufacturer operations manager, an MSP founder, and a Tier-1 technology provider video-analytics director.
- **Siemens partnership:** NodeWeaver is factory-installable on SIMATIC IPCs via ORCLA (single-click restore, no USB media).
- **Analyst recognition:** DCIG Top 5 SMB/Edge VMware vSphere alternative (2024–25); STL Partners 50 Edge Computing Companies to Watch in 2026; TM Forum Next20 winner, Intelligent Networks (2022); assessments by 451 Group, GigaOm, STL Partners, DCIG, and QKS Group.
- In production 11+ years, deployed in 15+ countries.

## Technology partners

Intel, HiveRadar, OnLogic, SuperMicro, Lenovo, Lanner, NetApp, CODESYS, WESCO, TD SYNNEX, Acronis, Mirasys, Accenture, Cloudasys, MergeIT.

## Company & values

Founded by infrastructure engineers who watched datacenter platforms break against the realities of edge deployment. Mission: minimize the lifecycle cost of the edge (management cost, acquisition cost, deployment cost).

Engineering principles: (1) The site is on its own — airgap and partition are default conditions. (2) No trained operators on site — if a procedure requires a skilled human at the edge, that's a defect. (3) Hardware-agnostic by force — heterogeneous fleets are the assumed default.

**Ethical Quality:** NodeWeaver's business approach — quality, fairness, and integrity compound. Pricing you can plan around, support engineers who know your environment, protected partner margins, a team treated fairly. "It's a deliberate way to build a company, and it's the only way we do business."

## Migrating from VMware

NodeWeaver is a DCIG Top 5 SMB/Edge VMware vSphere alternative (2024–25 report). Mapping: vSphere/ESXi → NodeWeaver virtualization; vSAN → Software-Defined Storage (full HA from 2 nodes); NSX → SDN & SASE; vCenter/Aria → EdgeInsight + DNSOps; Tanzu → built-in Kubernetes; per-core subscriptions → $1,500 perpetual per-node license. Details: https://nodeweaver.io/migrate-vmware.html · Migration guide: https://12.docs.nodeweaver.eu/#/migration_guide

## Contact

- Sales: sales@nodeweaver.io · https://nodeweaver.io/contact.html
- Engineering / architecture review (30 min, no obligation): engineering@nodeweaver.io · https://nodeweaver.io/book-meeting.html
- Support: https://ticket.nodeweaver.io/
- Press: press@nodeweaver.io
- Security reports: security@nodeweaver.io
- Privacy: privacy@nodeweaver.io
- Docs: https://docs.nodeweaver.io/
- LinkedIn: https://www.linkedin.com/company/nodeweaver/
- X / Twitter: https://x.com/nodeweaver