2020Summit-IoT-Streaming-Demo

Live Streaming and Analytics Demo for 2020 Virtual Summit

It is split up into three major pieces

IoT Camera Simulator -> Pulls video from any youtube livestream and pushes it to the Kafka http bridge
Analytics Service -> Gets live video data from kafka in the form of a Cloud Event, Sends the data to tensorflow-serving for inference, overlays the inference on the video, and uploads live video segments to ceph object storage
Video Serving Service -> Pulls the live video stream segments from ceph, builds the HLS playlist file, serves the inferenced video to the user with a static flask based webapp

The overall system diagram and main components are shown below

Deploy this demo yourself

This POC demo uses some great cloud-native technologies, follow the instructions to spin it up on a personal cluster It can be deployed with the manual instuctions shown below or the Simple IoTCLI instuctions if using the IoTCLI command line tool

IoTCLI Install

Prerequisites For IoTCLI Install

An Openshift Cluster 4.3>

For a quickstart Code ready containers may be used however they have not been tested with this system.

The latest IoTCLI download moved into yout $PATH

Provision Knative

Run IoTCLI knative setup which will setup knative eventing and serving on your cluster
Check status with IoTCLI knative setup --status Which should resemble the following

2020/04/24 15:14:00 knative Serving Install Status:
DependenciesInstalled==True
DeploymentsAvaliable==True
InstallSucceeded=True
Ready=True
2020/04/24 15:14:00 Context switched to namespace: knative-eventing
2020/04/24 15:14:00 NAME                                   READY   STATUS    RESTARTS   AGE
broker-controller-58fbf6569b-zgmz8     1/1     Running   0          12m
eventing-controller-5b9c5765dd-2w7cf   1/1     Running   2          12m
eventing-webhook-5cd6c688f4-7qzxw      1/1     Running   0          12m
imc-controller-845fc8776b-rcpzf        1/1     Running   0          12m
imc-dispatcher-5f9565cdbd-rmwlr        1/1     Running   0          12m

Provision Kafka

Run IoTCLI kafka setup to deploy kafka to your cluster with the strimzi Operator
Deploy the http->kafka bridge with IoTCLI kafka bridge
Follow the Instructions to check status

To check status of Kafka HTTP bridge follow the prompt at the end of the install logs 'curl -v GET http://<your bridge route>/healthy'

Which should show

< HTTP/1.1 200 OK
< content-length: 0
* Added cookie 9d04d3bd5bc37b277b860a02c6cf350d="69a5f362b53812ea1dbbee66d616958b" for domain my-bridge.io, path /, expire 0
< Set-Cookie: 9d04d3bd5bc37b277b860a02c6cf350d=69a5f362b53812ea1dbbee66d616958b; path=/; HttpOnly
< Cache-control: private
< 
* Connection #1 to host my-bridge.io left intact

Setup Ceph Object Storage

Run IoTCLI ceph setup to provision ceph

BE PATIENT This will take some time for all the resources to become available

Run IoTCLI ceph user <ceph username> to setup a new ceph user connected to credential store my-store
Run IoTCLI ceph secrets <ceph username> to get the login secrets and ceph endpoint URL for your ceph instance, these will be used later.

Setup the Tensorflow Serving Deployments

Run IoTCLI tensorflowServing setup -n kafka Which will spin up the tensorflow serving pod

Deploy the `Video Analytics` knative service

Run

IoTCLI knative service video-analytics -n kafka --cephEndpoint <Your Ceph Endpoint> --cephAccessKey <Your Ceph Access Key> --cephSecretKey<Your Ceph Secret Key>

Setup the Kafka -> Knative Bridge

Run

IoTCLI knative source kafka video-analytics -n kafka

Start the IoT Video Simulator

Navigate to the iotDeviceSimulator-kafka with

cd iotDeviceSimulator-kafka

Set the STREAMURL environment variable with

export STREAMURL=<Desired Youtube livestream>

Set Kafka Bridge Endpoint for this demo as follows

export ENDPOINT=<kafka-bridge-endpoint>/topics/my-topic

Start the Simulator

go run ./cmd

Deploy the `video-serving` service

Run

IoTCLI knative service video-serving -n kafka --cephEndpoint <Your Ceph Endpoint> --cephAccessKey <Your Ceph Access Key> --cephSecretKey<Your Ceph Secret Key>`

to deploy the service

Run IoTCLI service --status to get the url for the video serving service

NAME            URL                                                           LATESTCREATED         LATESTREADY           READY   REASON
video-analytics   http://video-analytics.kafka.apps.astoycos-ocp.shiftstack.com   video-analytics-895k8   video-analytics-895k8   True 
video-serving   http://video-serving.kafka.apps.astoycos-ocp.shiftstack.com   video-serving-9cfps   video-serving-9cfps   True

Follow the final step instructions

Manual Install from Source

Prerequisites For Manual Install from Source

An Openshift Cluster 4.1>

For a quickstart Code ready containers may be used however they have not been tested with this system.

Knative Eventing and Serving installed

Knative Operator install instructions
Knative Serving install instructions
Knative Eventing install instructions

Setup Kafka with some custom modules

Create a namespace for the Apache Kafka installation

oc create namespace kafka

Install the Strimzi operator

curl -L "https://github.com/strimzi/strimzi-kafka-operator/releases/download/0.16.2/strimzi-cluster-operator-0.16.2.yaml" \ | sed 's/namespace: .*/namespace: kafka/' \ | kubectl -n kafka apply -f -

Configure the Kafka Cluster, with custom max.message.bytes parameter

oc apply -n kafka -f demo_yamls/kafka.yaml

Setup HTTP-Kafka Bridge

oc apply -n kafka apply -f demo_yamls/kafka-bridge.yaml
Create a K8's Ingress Resource to access bridge from outside cluster
- kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/nginx-0.30.0/deploy/static/mandatory.yaml
- kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/nginx-0.30.0/deploy/static/provider/cloud-generic.yaml
- oc apply -f demo/yamls/ingress.yaml

Run curl -v GET http://my-bridge.io/healthy to check bridge, output should be as follows:

> GET /healthy HTTP/1.1
> Host: my-bridge.io:80
> User-Agent: curl/7.61.1
> Accept: */*
> 
< HTTP/1.1 200 OK
< content-length: 0

Setup Rook and Ceph Object storage

Follow Open Data Hub's instructions for Ceph installation with the Rook operator

And make sure to save the resulting s3 credentials

Deploy Tensorflow Serving

Do to the video inference the demo using a tensorflow-serving container with a pre-trained object detction model from the Model zoo library specifially using the ssd_inception_v2_cocossd_inception_v2_coco Model

To deploy this model simply run

oc apply -n kafka -f demo_yamls/tensorflow-deployment.yaml

[astoycos@localhost 2020Summit-IoT-Streaming-Demo]$ oc get pods --selector=app=coco-server
NAME                                    READY   STATUS    RESTARTS   AGE
tensorflow-deployment-9d867d795-5q4kb   1/1     Running   0          2d4h

To Ensure the Serving pod is up run

oc get pods --selector=app=coco-server It should show the following

To get the podIP that will be used by the Knative Video Service Run:

export IP=$(oc get pods --selector=app=coco-server -o jsonpath='{.items[*].status.podIP'})

Deploy Knative Video Service

Apply your S3 Credentials and endpoint in demo_yamls/video-service.yaml.in
Apply the knative service with the environment configs run

cat demo_yamls/video-service.yaml.in | envsubst | oc apply -n kafka -f -

Make sure the knative service is ready with

oc get ksvc Which should resemble the following

NAME            URL                                                           LATESTCREATED         LATESTREADY           READY   REASON
video-service   http://video-service.kafka.apps.astoycos-ocp.shiftstack.com   video-service-895k8   video-service-895k8   True

Deploy the Kafka-> Knative Source with

oc apply -n kafka -f demo_yamls/kafka-event-source.yaml

Now the Analyzed video stream segments should be stored in Ceph

Start the IoT camera simulator

Navigate to the iotDeviceSimulator-kafka with

cd iotDeviceSimulator-kafka

Set the STREAMURL environment variable with

export STREAMURL=<Desired Youtube livestream>

Set Kafka Bridge Endpoint for this demo as follows

export ENDPOINT=my-bridge.io/topics/my-topic

Start the Simulator

go run ./cmd

Deploy Knative Serving Service

Apply your S3 Credentials and endpoint in demo_yamls/video-serving-service.yaml
Apply the service with

oc apply -n kafka demo_yamls/video-serving-service.yaml

Get the service URL with

oc get ksvc

NAME            URL                                                           LATESTCREATED         LATESTREADY           READY   REASON
video-serving   http://video-serving.kafka.apps.astoycos-ocp.shiftstack.com   video-serving-jsct5   video-serving-jsct5   True

Move to Final Step instructions

Final Steps

Go to the a web browser and type in the video-serving service's URL. Add /video/out.m3u8 to the path for a final URL as follows

http://video-serving.kafka.apps.astoycos-ocp.shiftstack.com/video/out.m3u8

The following webpage will resemble the following, showing live video streaming an anlytics

Note:: You may need to add the service's URL to your etc/hosts file for correct DNS lookup

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

README.md

README.md

2020Summit-IoT-Streaming-Demo

Deploy this demo yourself

IoTCLI Install

Prerequisites For IoTCLI Install

Provision Knative

Provision Kafka

Setup Ceph Object Storage

Setup the Tensorflow Serving Deployments

Deploy the `Video Analytics` knative service

Setup the Kafka -> Knative Bridge

Start the IoT Video Simulator

Deploy the `video-serving` service

Manual Install from Source

Prerequisites For Manual Install from Source

Setup Kafka with some custom modules

Setup Rook and Ceph Object storage

Deploy Tensorflow Serving

Deploy Knative Video Service

Start the IoT camera simulator

Deploy Knative Serving Service

Final Steps

Files

README.md

Latest commit

History

README.md

File metadata and controls

2020Summit-IoT-Streaming-Demo

Deploy this demo yourself

IoTCLI Install

Prerequisites For IoTCLI Install

Provision Knative

Provision Kafka

Setup Ceph Object Storage

Setup the Tensorflow Serving Deployments

Deploy the Video Analytics knative service

Setup the Kafka -> Knative Bridge

Start the IoT Video Simulator

Deploy the video-serving service

Manual Install from Source

Prerequisites For Manual Install from Source

Setup Kafka with some custom modules

Setup Rook and Ceph Object storage

Deploy Tensorflow Serving

Deploy Knative Video Service

Start the IoT camera simulator

Deploy Knative Serving Service

Final Steps

Deploy the `Video Analytics` knative service

Deploy the `video-serving` service