Skip to main content
Version: v3.9.3

Core Concepts

The enterprise edge has become a business-critical frontier. Today, up to 75% of enterprise data is created outside the data centers. It is generated across factories, stores, remote sites, vehicles, and infrastructure. The scale is massive.

At the same time, edge environments remain resource-constrained by design. They operate in real-world conditions, with limited connectivity and tight operational margins. Yet the hardware landscape is changing fast. New generations of edge CPUs and GPUs are now powerful, affordable, and widely available. This makes real-time analytics, Edge AI, and autonomous decision-making practical at the edge.

Traditional centralized cloud models cannot keep up. Latency is dictated by physics. Costs grow with every round trip. When connectivity is degraded, remote sites lose autonomy. That creates risk, complexity, and operational friction for distributed deployments.

Eclipse ioFog brings cloud-native principles to the edge without ignoring real-world constraints. It is secure, vendor-agnostic, and open by design, reducing operational complexity while preserving local autonomy at each site.

Introducing Eclipse ioFog​

Eclipse ioFog onboards edge sites as Edgelet nodes. Edgelet is the node agent, the application runtime, and, on Linux, the container engine in one binary. The Linux OTA download is about 30 MB and already contains the engine, so a host does not need Docker or Podman. Linux builds cover amd64, arm64, arm (32-bit), and riscv64. The same binary can run standalone from the edgelet CLI, then join a cluster with iofogctl.

Installation and upgrades go through iofogctl and the EdgeOps Console. The platform picture is on Architecture. The engine itself is Get to know Edgelet.

Key capabilities​

Eclipse ioFog provides the capabilities required to run and operate software workloads across distributed edge environments. It focuses on the full lifecycle of applications, services, and edge operations with unified control at scale.

Edge environment and Edgelet nodes​

Eclipse ioFog simplifies how edge environments are onboarded. Edgelet nodes use the Edgelet runtime to establish secure identity and connectivity to the Control Plane. Installation and upgrades are automated through iofogctl and the EdgeOps Console, without deep per-site customization.

The platform provides an open operational layer above hardware and operating systems. You manage diverse environments as one cluster while preserving local isolation.

Application and workload lifecycle orchestration​

Eclipse ioFog orchestrates the complete lifecycle of containerized workloads at the edge. Applications are deployed, updated, versioned, and retired consistently using declarative YAML manifests.

Environmental differences are abstracted so teams focus on what should run and where, not on per-site wiring. That enables repeatable deployments, controlled rollouts, and safe updates across edge clusters.

Configuration, secrets, and policy management​

Configuration values and secrets are defined centrally and distributed securely to edge workloads at runtime. Role-based access control (RBAC) ensures users, automation, and workloads stay within intended scope.

AI models and knowledge assets​

Platform train v3.9.0 adds fleet catalogs for Edge AI assets using the same declarative model as applications and registries.

  • Model catalog (kind: Model): LLM and inference weights. Sources include Hugging Face Hub models and OCI artifacts (model-spec, ModelPack, ModelKit, and ORAS-compatible blobs) through registry type hf or oci.
  • Knowledge catalog (kind: Knowledge): RAG corpora, datasets, and document bundles. Hugging Face datasets and OCI bundles use the same registry types. Weights stay in Model; knowledge uses a separate namespace.

Typical lifecycle:

  1. Define Model or Knowledge YAML and apply it to the Controller catalog.
  2. Attach assets to one or more Edgelet nodes so the platform pulls and tracks the right revision.
  3. Bind catalog names in microservice spec.models and spec.knowledge so running workloads use fleet-managed content, not host paths.

EdgeOps Console exposes Config → AI Model Catalog and Config → AI Knowledge Catalog. Operators can also use iofogctl deploy, attach model, attach knowledge, and matching describe commands.

Learn more: AI Model Catalog, AI Knowledge Catalog, Microservices - models and knowledge, and the Edge AI wafer defect tutorial.

Secure connectivity​

Eclipse ioFog establishes trust between the Control Plane and edge devices. Each Edgelet node and workload uses verified identity with built-in certificate management.

A secure service mesh (Router, Skupper-based) and distributed message bus (NATS) are provided by default. Services communicate with encryption across nodes and sites while edge networks can remain isolated from a routing perspective.

Observability and operations​

Teams observe workload status, connectivity, and health across sites from one operational view. iofogctl supports automation-first workflows. The EdgeOps Console supports day-two operations, including v3.9 Configure flows for AI catalogs and cluster config.

Ready for more?
Group 3See anything wrong with the document? Help us improve it!

Edit this page on GitHub!