Platform

A complete edge stack — engineered for the complexity of real deployment.

NodeWeaver combines virtualization, software-defined storage and networking, autonomous orchestration, integrated SASE, and continuous security — running on your choice of standard x86 hardware. One stack, any hardware, anywhere.

01 / Virtualization

Efficient by design. Tunable when it matters.

Everything runs in either 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.

  • Automated placement across heterogeneous hardware
  • Legacy Windows workloads in VMs alongside modern containers
  • Continuous tuning of OS & hypervisor parameters
  • Manual pinning, NUMA topology, affinity / anti-affinity rules
  • PCI passthrough and network card passthrough
  • Live hardware replacement, no downtime
AUTOMATED PLACEMENT Orchestrator · scheduler placement · pinning · continuous tuning host-01 4c · 8G VM thinVM host-02 24c · 64G VM VM RT·pinned K8s thinVM host-03 8c · 16G VM thinVM HETEROGENEOUS BARE-METAL HARDWARE VM / K8s thinVM RT soft-realtime, pinned
Ubuntu — idle~ 5–25 μs · spikes > 100 μs
Ubuntu — CPU stress200 μs – 10 ms tail
NodeWeaver realtime VM — idleP99.99 < 50 μs
NodeWeaver realtime VM — saturatedP99.99 < 150 μs
0 2.5 ms 5 ms 7.5 ms 10 ms
02 / Soft-realtime

Bounded latency, even under full CPU saturation.

Latency and jitter are tunable through a progressive set of attributes — REALTIME, CPU_PARTITION, pinned tasks, custom NUMA topology, hypervisor attributes, and 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 in our soft-RR benchmark.

NodeWeaver provides soft real-time guarantees for supervisory, inspection, and HMI workloads.

03 / Software-Defined Storage

Distributed storage that survives the site.

A built-in distributed filesystem, transparent to workloads. Chunk servers encrypt data at rest. Snapshots, backups and replication are first-class. The cluster keeps storage available through disk loss, node loss and full power events.

  • Distributed chunkserver engine, per-tenant encryption
  • Native snapshots, backup & replication
  • Survives disk loss, node loss, full power event
  • Benchmarked competitively against Longhorn at 3-node CSI
DFS · 2-NODE HA · GROW ONE NODE AT A TIME VMs · K8s · Appliances · App data Distributed FS · master · metalogger replicated · encrypted · snapshot-capable node-01 encrypted data encrypted data node-02 encrypted data encrypted data 2-NODE HA +1 node encrypted data transparently add capacity +N ···
Central Management / Monitoring EdgeInsight · DNSops · Terraform · Ansible site · maritime control plane 3 nodes · self-run independent site · tower-12 control plane 7 nodes · RT independent site · solar control plane 2 nodes · offline keeps running EACH CLUSTER RUNS ITS OWN CONTROL PLANE · MANAGE CENTRALLY WHEN CONNECTED
04 / Software-Defined Network & SASE

One encrypted plane to every node. Forever.

Software-defined networking 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 when uplinks are flaky or asymmetric.

  • SASE integrated — no third-party VPN to manage
  • Distributed firewall, DDoS protection
  • Wire-line speed for Ethernet up to 100 Gbit/s
  • Works in limited connectivity and airgapped environments
  • Fleet management traffic under 1 MB/hour — or zero, fully airgapped
05 / Security

Security as a continuous property — not a checklist.

NodeWeaver treats security as a Day-N concern, not a Day-0 launch task. Encrypted, tamper-proof firmware on every device. Small incremental μ-updates that don't require reboots — and that can ship to airgapped clusters.

  • Encrypted, signed, tamper-proof firmware
  • Encryption-at-rest for all stored data
  • Hardened OS, hardened management plane
  • Strong workload insulation, resource reservation
  • Tamper detection and remote wipe
  • μ-updates with no reboot — even when airgapped
  • Full logging, traceable APIs, auditable end-to-end
  • Designed in alignment with IEC 62443
DEFENSE IN DEPTH SIGNED, TAMPER-PROOF FIRMWARE ENCRYPTION-AT-REST ENCRYPTED MGMT · DDoS WORKLOAD ISOLATION AUDIT · μ-UPDATES continuous · no reboot airgap-capable auditable
FLEET CONTROL PLANE One declarative API GitOps · self-heal · update windows · policy region · us 142 sites · in sync healthy region · eu 96 sites · 1 remediating auto-healing region · apac 74 sites · in sync healthy 312 sites · 1,486 nodes · one console in sync drift · auto-remediating
06 / Orchestration

Fleet-scale ops. Customer self-service.

One declarative API governs every cluster from one site to one thousand. Self-heal, self-remediate, GitOps-native, fully scriptable. Customers onboard themselves — your engineers don't need to be in the loop.

  • Per-node licensing, flexible scaling
  • Customer self-service onboarding
  • Autonomous remediation, fleet-wide
  • Open REST APIs, with Ansible and Terraform support
  • Checkmk export and rsyslog log streaming — feeds Splunk, Datadog, Grafana, or any SIEM
  • Integrates with Azure Arc and AWS SSM
See the platform

The actual product — not a mockup.

NodeWeaver's built-in web UI and the EdgeInsight monitoring console, captured from a production fleet.

NodeWeaver web UI — cluster dashboard showing VMs, images, virtual networks, host health, and full-HA status
NODEWEAVER WEB UI · cluster dashboard — VMs, storage, networks, host health, live HA status
EdgeInsight monitoring — all hosts across US and EU infrastructure
EDGEINSIGHT · fleet view — every host, every region, one console
EdgeInsight monitoring — per-service checks of a single host with performance meters
EDGEINSIGHT · host detail — per-service checks, latencies, uptime
Use cases

Six places we live in production today.

From maritime fleets to wind farms to factory floors — NodeWeaver is in the cabinets you don't see, doing the work that doesn't fit in a datacenter.

Surveillance

Real-time AI video

Casinos, municipalities, datacenters running on-prem inference on security cameras.

Edge NFV

SD-WAN & packet

Distributed network function virtualization with HA at 14K+ IP endpoints.

uCPE

VM & K8s in 4c / 8GB

API-driven workload placement on tight-footprint universal premises hardware.

Distributed Edge

5G RAN & CDN

Bounded-latency processing for radio stacks, gaming, video, content delivery.

Energy

Wind & solar farms

Intelligent power management in remote, intermittently-connected installations.

Industrial

Manufacturing

Real-time weld inspection, factory automation, IoT, 3D-printing for medical devices.

Specifications

What you need to know before procurement.

NodeWeaver runs on commodity x86 hardware. Below are some of the most common questions, please reach out if you'd like to learn more.

Minimum node
1 core · 1 GB RAM

For uCPE: 4 cores / 8 GB recommended with HA.

Cluster size
1 → N nodes

No theoretical ceiling — limited by your network plane.

Realtime guarantee
P99.99 < 150 μs

With REALTIME + CPU_PARTITION. 12 samples / 1 billion outside bound.

Throughput
Up to 100 Gbit/s

Wire-line speed for Ethernet.

Deploy time
< 10 minutes

From drop-shipped box to operative nanocloud. No human on site.

Update model
μ-updates, no reboot

Including airgap clusters. Reboot-needed updates are explicitly tagged.

Hardware
x86 · any vendor

Hardware-agnostic. Pre-validated with SuperMicro, Lenovo, Dell, HP.

Node roles
All nodes equal

Every node runs the same software. No dedicated management or control-plane node to size, license, or lose.

Schedule a no-obligation architecture review with a NodeWeaver Engineer.

Book the review See pricing