Edge Benchmarking Suite

A small collection of Edge Computing benchmarking tools

How to use

This repository describes setting up three open source Edge Platforms, and includes sample configurations for reading, filtering, and writing to MQTT.

The MQTT messages are timestamped by each stage in their processing, and the difference in the timestamps are visualized with the help of Node-Red.

Below are detailed instructions for setting up the messaging brokers, and each of the platforms evaluated.

This repository includes:

• A configurable MQTT Random Number Generator
• Configurations for Eclipse Kura
• Configurations for Linux Foundation EdgeXFoundry
• Configurations for TIBCO Flogo

All scripts in the root of this repository are intended for running on ARM architecture (RPi or other) Some may be appliciable to other architectures. See /x86/

Tools Used: MQTT messaging service (Mosquitto) node-red (flow creation) node-red dashboard (visualization) time POSIX terminal applicatoin ctop Container resource monitor

Starting MQTT messaging service

snap install mosquitto mosquitto-clients

-or as a docker image-

docker run -d --rm --name broker -p 1883:1883 eclipse-mosquitto

Starting virtual MQTT device

note:

• replace the below IP with your device IP from ifconfig
• replace path where you checked out this repo: -v /home/pi/edge-metrics/vir-mqtt-dev

on your Raspberry Pi communication node:

docker run -d --rm --name=mqtt-scripts \
-v /home/ubuntu/edge-metrics/vir-mqtt-dev:/scripts \
dersimn/mqtt-scripts:armhf --url mqtt://192.168.1.1 --dir /scripts

• Or on a x86 machine:

docker run -d --rm --name=mqtt-scripts \
-v /home/dsouthwi/edge-metrics/vir-mqtt-dev:/scripts  \
dersimn/mqtt-scripts --url mqtt://128.141.207.157 --dir /scripts

note: if you get problems mounting the vir-mqtt-dev folder, try disabling selinux (check docker logs --latest)

Verify readings are being sent to the MQTT broker

Many ways to do this; I chose to use the mosquitto client (snap install mosquitto) Here, we subscribe to the topic declared by our virtual device in mqtt/configuration.toml

You should see output similar to this after 15 seconds (default interval)

# mosquitto_sub -v -t '#'
logic/connected 2
DataTopic {"name":"MQ_DEVICE","cmd":"randnum","randnum":"25.6"}
DataTopic {"name":"MQ_DEVICE","cmd":"randnum","randnum":"28.3"}

Start node-red

nodered-start and import the included nodered-flow.json for visualizing results

EdgeX

Configuration Installation

Install helper packages used in this repo for lua-based API scripts:

sudo apt-get install lua5.1 lua-socket lua-dkjson

(note any version of Lua will work, the script shebang may need to be updated)

Starting EdgeX

Use the docker-compose.yml file in the base of this repo:

docker-compose pull
docker-compose up -d

####### Docker x86

Pull latest docker-compose file

wget https://github.com/edgexfoundry/developer-scripts/raw/master/releases/edinburgh/compose-files/docker-compose-edinburgh-1.0.1.yml -O docker-compose.yml

and compose docker-compose pull docker-compose up -d

Configuring edgex

The below steps are already included in the above rpi docker-compose, and helper script

EdgeX is service based, and actions need to be defined at each service. First, define the MQTT service and behaviour: Using device-mqtt service block of docker-compose.yml

  device-mqtt:
    image: edgexfoundry/docker-device-mqtt-go:1.0.0
    ports:
      - "49982:49982"
    container_name: edgex-device-mqtt
    hostname: edgex-device-mqtt
    networks:
      edgex-network:
        aliases:
          - edgex-device-mqtt
    volumes:
      - db-data:/data/db
      - log-data:/edgex/logs
      - consul-config:/consul/config
      - consul-data:/consul/data
      - ./mqtt:/custom-config
    depends_on:
      - data
      - command
    entrypoint:
      - /device-mqtt
      - --registry=consul://edgex-core-consul:8500
      - --confdir=/custom-config

start device-mqtt docker-compose up -d device-mqtt

Add publishing behaviour as defined in EdgeX-register-MQTT-publish.lua: ./Edgex-register-MQTT-publish.lua 192.168.1.1 where 192.168.1.1 is the location of your MQTT server to publish to.

Return codes from the command should be the UUID of the new export registration: 1 146c578b-d5b9-4c39-9e36-65e7a1f4f1d3

EdgeX stores persistent data in docker volumes. To clear this data, delete the volumes: Starting from a clean slate: docker system prune --volumes

Kura

Installation

https://eclipse.github.io/kura/intro/raspberry-pi-quick-start.html

Alternatively, Kura is available from Dockerhub: docker run -d -p 8080:8080 -t eclipse/kura

Adding kura wires script block from the eclipse store (drag and drop in Kura 'Packages' panel) https://marketplace.eclipse.org/content/wires-script-filter-kura-4xy

Then import the included kura configurations kura_settings.xml into the Kura 'Settings' panel

Testing the included kura topic: mosquitto_pub -t "kura/02:42:AC:11:00:03/W1/DataTopic" -m "{ \"sentOn\":1511970625597, \"metrics\":{\"fieldName\":12345},\"body\":12344555}"

Flogo

Install golang, apt install golang export go settings:

export GOPATH=$HOME/go
export PATH=$PATH:$GOPATH/bin
export GOPROXY=direct

install Flogo CLI go get -u github.com/project-flogo/cli/... Build the flogo.json FLOW in this repository

flogo create -f edge.json edge
cd edge
flogo build -e

Deploy the resulting binary in edge/bin to your edge device and execute.