CIOPages
DirectoryAkri

Akri

Open Source

Kubernetes interface for dynamic edge device discovery and utilization

Visit Website

About Akri

Akri is an open-source Kubernetes Resource Interface designed to expose heterogeneous edge devices such as IP cameras, USB devices, GPUs, and FPGAs as consumable resources within Kubernetes clusters. It addresses the challenge of integrating diverse and intermittently available leaf devices that are too small to run Kubernetes themselves, enabling seamless edge computing deployments. By leveraging and extending the Kubernetes device plugin framework, Akri dynamically discovers devices using configurable Discovery Handlers and schedules workloads accordingly.

Targeted at enterprises adopting Kubernetes for edge computing, Akri simplifies the management and utilization of a wide array of embedded hardware resources. It abstracts the complexity of device discovery, availability monitoring, and workload scheduling, allowing operators to deploy configurations that automatically detect devices and launch associated pods. Akri also supports high availability by enabling multiple nodes to access the same device and automatically creates Kubernetes services for device types, reducing operational overhead. Its extensible architecture supports multiple discovery protocols and encourages community-driven expansion, making it a flexible solution for evolving edge environments.

Key Capabilities

  • Dynamic discovery of heterogeneous edge devices
  • Integration with Kubernetes device plugin framework
  • Support for multiple discovery protocols (ONVIF, udev, OPC UA)
  • Automatic workload scheduling based on device availability
  • High availability through multi-node device access

Integrations

KubernetesPrometheusCloud Native Computing Foundation (CNCF) ecosystem

This profile was compiled by CIOPages from public sources with AI assistance, and may be incomplete or out of date. It is informational only and not an endorsement. Represent this vendor? or .

Quick Facts

docs.akri.sh
PricingOpen Source
DeploymentOpen Source
Target SizeEnterprise