Skip to content

Latest commit

 

History

History
108 lines (66 loc) · 11.5 KB

README.md

File metadata and controls

108 lines (66 loc) · 11.5 KB
Raw

Setting up Clound Native Edge systems with the IoTCLI

The user can choose which tools they want to use in order to either digest, display or process the IoT data. Eventually the user will be allowed to create custom data pipelines to connect the tools. Currently the following tools/commands are supported

Getting Started

Download the latest Release from the release page

Move the Binary IoTCLI into your $PATH

Run IoTCLI login to authenticate with your Openshift cluster

Main Components Overview

Messaging Core

In any Edge system that incorporates IoT devices, having a scalable cloud native edge system is very important. With the IoTCLI you can easily deploy these messaging services to your Openshift 4.X cluster

Enmasse

Enmasse is an open source project for managed, self-service messaging on Kubernetes.

The IoTCLI makes it easy to deploy the system along with its IoT services onto an openshift Cluster

Kafka

Apache Kafka is an advanced distributed streaming platform. It is deployed to your Openshift cluster using the Strimzi operator

Application Core

Once IoT devices are streaming or pinging data from the edge developers need a way to process the data either on the Edge or back in the central cloud cloud

In-Cloud

Knative

The IoTCLI can install and setup Knative on an Openshift cluster, deploy Knative Services, and Container Sources

Currently configured Knative Services

event-display

  • Simply Drops incoming Cloud-Events into its log

iotVideo

  • Accepts an incoming IoT video Livestream as HLS Playlist File(.m3u8), runs image classification using Tensorflow, and serves it back to the user via a simple web appliaction, its repo can be found here

video-analytics

  • Accepts a video livestream in form of an HLS Playlist file delivered via a JSON structured to work with kafka. Sends video to Tensorflow serving module for inference, and sends inferenced video segments ceph object storage for eventual serving. Part of 2020 Summit Demo

video-serving

  • Serves analyzed live video from ceph to user via a simple flask web app. Part of 2020 Summit Demo

Currently configured Knative Sources

iot

  • AMQP-> CloudEvent broker to ferry messages from the application side of Enmasse to a Knative Service, its repo can be found here

kafka

  • KafkaJSON -> Cloud Event broker to ferry video playlist files from Kafka to Knative

Persistent Storage

Persistent storage is implemented using Ceph Object Storage deployed via the Rook Operator V1.3.2

On Edge

Gstreamer

Gstreamer is an open source multimedia framework for managing media files and real time streaming using camera. This project leverages gstreamer capabilities with CNN model based enchanced analytics from OpenVino toolkit by Intel. This can be used to perform HLS video streaming with various use cases such as object detection, classification, recognition and tracking.

TensorFlow lite

TODO

Command Reference

Command1 (Tool) Command2(Module) Command3(Module command) Argument (module input) Flags Function
ceph destroy None None None Destroy the ceph object store instance on the cluster
ceph user None <User Name> None Create an Ceph Object Store User
ceph secrets None <User name> None Return user secrets for ceph deployment
ceph setup None None None Setup Ceph Object Storage via the Rook Operator
enmasse destroy None None None Remove Enmasse from openshift cluster
enmasse IoT addDevice <Messaging Tenant>, <DeviceID> None Add a Device with specified ID to the Enmasse device registry for a specified messaging TenantSetup default Credentials
enmasse IoT project None --namespace Make a new enmasse IoT project in the specified namespace, defaults to “myapp”
enmasse setup None None None Download Enmasse Source, store in current directory. Setup Enmasse Setup IoT services
gstreamer setup None None --local=true/false Deploy Gtsreamer with OpenVino Toolkit for video/audio streaming with analyticsSet local=True for deploying gstreamer locally
gstreamer destroy None None None Destroy Gstreamer deployment
kafka bridge None None --namespace --route=true/false Deploy kafka HTTP bridge Set route=True to deploy a route to access bridge from outside(only for Openshift Cluster)Default deploys nginx ingress to access bridge from outside the cluster
kafka destroy None None --namespace Destroy the kafka deployment located at the specified namespace
kafka setup None None --namespace Setup a kafka cluster viat the strimzi operator at the specified namespace
knative setup None None --status=true/false Setup Knative serverless on openshift clusterConfigures both Knative-Eventing and Knative-ServingSet --status=true to check on Knative deployment
knative destroy None None None Remove Knative deployment from openshift cluster
knative service None <Knative service to be deployed> --status=true/false--namespace Deploy a knative service Set --status=true to check on Knative service deploymentDefault namespace is “knative-eventing”
knative service destroy <Knative service to be destroyed> --namespace Remove a specified Knative service from the cluster at specified namespace Default namespace is “knative-eventing”
knative source None <containersource to be deployed> --namespace Deploy a Knative Source at specified namespace Defaults to namespace “knative-eventing”
knative source destroy <containersource to be destroyed> --namespace Remove a specified knative source from the cluster from specified namespaceDefault namespace is “knative-eventing”
login None None None None Login to your openshift cluster with the username and password
tensorflowServing setup None None --namespace Deploy a tensorflow serving deployment and service Default namespace is “default”
tensorflowServing destroy None None --namesapce Deploy a tensorflow serving deployment and serviceDefault namespace is “default”