Now we need to setup the Azure side and for that you need an Azure subscription. You can get a new subscription with 170€ of credits for 30 days with this link, if you don’t have anything yet:
As this is a demo environment, we do not focus heavily on security. You need some basic understanding how to deploy resources. I’ll guide you on the configuration side. Btw, If you want to learn basics of Azure, there is a nice learning path at Microsoft Learning:
Here you can see my demo environment:
IoT Hub is for receiving temperature messages from Raspberry Zero. Stream Analytics Job is for pushing those messages to the Power BI. The Automation Account is for starting and shutting down the Stream Analytics job, so it won’t consume too much credits.
Normally, we would deploy this process with ARM, but for make it more convenience to present and instruct, let’s do it from the Portal. And if you like, you can follow naming convention from CAF:
Start deploying the IoT Hub. Use a globally unique name and public endpoints (You can restrict access to your home ip-address if you like) and choose F1: Free Tier. With the Free Tier you can send 8000 messages per day, so if you have five Raspberry Pis, each of them can send one message per minute. Enough for home use usually.
After you have created the IoT Hub, you need to create an IoT Device under it. I used same hostname as my Raspberry Pi has:
Then you need to copy your Primary Connection String from your IoT Device:
After copying that string, you need to create an environment variable for your Raspberry Pi. You can use a script to automatically add it after every boot.
Here is an example how you create an environment variable:
Last thing to IoT Hub is to add a consumer group. You can add it under Built-in Endpoints and Events. Just add a custom name under $Default. Here you can see, that I added ‘ruuvitagcg’:
Next, you want to create a Stream Analytics Job. For that, you need only one unit and environment should be Cloud. There is no free tier for this and it costs some money to keep it running. Luckily, we can turn it off whenever we don’t use it. I used the Automation Account to start it a few minutes before I receive a message and turn it off few minutes after. There is a minor cost for the Automation Account also, but without it, the total cost would be much higher. I receive a message only once per hour, so Stream Analytics is running only 96 minutes every day, instead of 1440 minutes. The total monthly cost is something like 4€. Normally it would be almost 70€.
Here are my Automation Account scripts:
$connectionName = "RuuvitagConnection" $servicePrincipalConnection=Get-AutomationConnection -Name $connectionName Connect-AzAccount ` -ServicePrincipal ` -TenantId $servicePrincipalConnection.TenantId ` -ApplicationId $servicePrincipalConnection.ApplicationId ` -CertificateThumbprint $servicePrincipalConnection.CertificateThumbprint Start-AzStreamAnalyticsJob -ResourceGroupName "YourRG" -Name "YourStreamAnalytics" -OutputStartMode "JobStartTime"
$connectionName = "RuuvitagConnection" $servicePrincipalConnection=Get-AutomationConnection -Name $connectionName Connect-AzAccount ` -ServicePrincipal ` -TenantId $servicePrincipalConnection.TenantId ` -ApplicationId $servicePrincipalConnection.ApplicationId ` -CertificateThumbprint $servicePrincipalConnection.CertificateThumbprint Stop-AzStreamAnalyticsJob -ResourceGroupName "YourRG" -Name "YourStreamAnalytics"
Next, head to Stream Analytics Job and click Inputs.
Click Add stream input. Above is my configuration, you just need to add your own consumer group configured earlier. For Endpoint, choose Messaging. Use service for Shared access policy name (It is created default with a new IoT Hub).
Now, move on to Outputs and click Add. Choose Power Bi and click Authorize, if you dont have Power BI yet, you can Sign Up.
Fill in Dataset name and Table name. For the Authentication mode, we need to use User token if you have a Free Power BI version (v2 upgrade is not yet possible in Free).
Then create the following Query and click Test query, you should see some results:
Now, the only thing to do is to visualize our data with the Power BI. We will cover that part in the next post, but the infrastructure side is ready to rock. Grab a cup of coffee and pat yourself to the back 🙂