Integrating SQLite in Azure IoT Edge for local storage

Microsoft has opened the Azure Marketplace for IoT Edge modules. This makes it easy to find and install IoT Edge modules into your own IoT Edge devices.

At this moment a dozen or more modules are available:

We see a few well-known modules from Microsoft which support SQL Server, Modbus, OPC-UA or the Temperature sensor simulation.

One of these modules is supporting SQLite.

What is SQLite?

According to the homepage, SQLite is an in-process library that implements a self-contained, serverless, zero-configuration, transactional SQL database engine. The code for SQLite is in the public domain and is thus free for use for any purpose, commercial or private. SQLite is the most widely deployed database in the world with more applications than we can count, including several high-profile projects.

What does the module do?

This IoT Edge module is capable of accessing SQLite databases locally.

Microsoft says:

“Using this module, developers can build Azure IoT Edge solutions with the capability to access SQLite databases. The SQLite module is an Azure IoT Edge module, capable of executing SQL queries sent from other modules, and return the result to the senders or to the Azure IoT Hub via the Edge framework. Developers can modify the module tailoring to any scenario.”

The module is open source and both documentation and source code are put on GitHub.

Why should we use this module?

In the past, I checked out the SQL Server IoT module to persist IoT messages locally.

This SQLite module has a few advantages which you will like:

• It’s lightweight
• It’s based on a public domain framework
• It’s integrated with the routing of IoT Edge

For me, that last item, we can integrate the module directly in the routing is interested.

So let’s check out how we can use it.

Advertenties

Custom IoT Hub assignment in Device Provisioning Service

In my previous blog, I have shown how to provision a device using a real TPM using the Device Provisioning Service (DPS).

Once you are able to provision your IoT devices to the Azure IoT Platform using a DPS, a whole new world of possibilities opens up for you.

Before, you registered your device to one IoTHub. To change it, you had to go to the device and fix it. But now you are able to make a choice between multiple IoT Hubs within the cloud, dynamically!

But what strategy are you going to use?

Microsoft provides three standard strategies out of the box:

1. Lowest latency (select the nearest IoT Hub)
2. Evenly weighted distribution (select the IoT Hub with the least amount of devices)
3. Static configuration (just select one yourself. This is the situation as before)

But there is a new strategy which is very flexible:

This fourth strategy makes use of a custom Azure Function which you can write yourself.

You could, for instance, access a database and read some data before you make the decision to which IoTHub you assign this device.

Let’s see how we can build a custom function ourselves and get the most out of it.

Provision your IoT Edge device using a TPM

IoT Devices need a secure connection, the reason is obvious. It all starts with a secure connection between a device in the field and the Cloud platform.

Microsoft provides a secure connection for devices to the IoT Hub in three ways:

• Symmetric keys
• Certificates
• Support for a Trusted Platform Module (TPM)

We are interested in how to get our connection secured using the TPM.

A TPM is “an international standard for a secure cryptoprocessor”. It can generate private keys and expose the public keys related to them. So it somewhat behaves like a set of certificates but now as a physical device.

If somebody tries to physically compromise the chip to retrieve a private key, it should break thus destroying the chip and its content. You can buy TPM chips (eg. for a Raspberry PI) but it’s better to have it already attached to your PC’s motherboard. The chip acts as an identity and you do not want to see it being unplugged.

My Advantech UNO 2372G has a TPM 2.0 chip already built in. The same goes for a few laptops I have. Keep in mind that older versions of this security chip (like the TPM 1.2) are not supported by Microsoft.

Symmetric keys and certificates are supported by the Azure IoT Hub. We need another service, the Azure Device Provisioning Service to provision a device using the TPM and get access to an IoT Hub of your choice.

How does it work?

The Device Provisioning Service acts as a broker between provisioned devices and one or more IoTHubs.

The following picture shows the ‘dance’ a registered device has to perform if it wants to contact an IoT Hub (example taken from the documentation):

1. Device manufacturer adds the device registration information to the enrollment list in the Azure portal.
2. Device contacts the provisioning service endpoint set at the factory. The device passes the identifying information to the provisioning service to prove its identity.
3. The provisioning service validates the identity of the device by validating the registration ID and key against the enrollment list entry using either a nonce challenge (Trusted Platform Module) or standard X.509 verification (X.509).
4. The provisioning service registers the device with an IoT hub and populates the device’s desired twin state.
5. The IoT hub returns device ID information to the provisioning service.
6. The provisioning service returns the IoT hub connection information to the device. The device can now start sending data directly to the IoT hub.
7. The device connects to IoT hub.
8. The device gets the desired state from its device twin in IoT hub.

Note: Keep in mind, this dance can only start after a device is registered at the DPS. There must be a trust relationship between the device (with a TPM) and Device Provisioning Service first.

Why should we use a TPM?

Every example starts with symmetric keys. Are symmetric keys not enough?

All three available ways to secure a device are great but only certificates and a TPM are recommended to be used in production. The problem with symmetric keys is that replacing those keys is hard, you need to change it on the device itself. And you need to transport the new key to the device (on a USB stick?) so you are a bit vulnerable then.

Using a TPM (and a DPS) helps in two ways:

1. When the security token behind the secure connection with an IoT Hub expires, the device itself simple asks for a new token by connecting to the DPS. There is no need for extra work to be done.
2. The DPS has knowledge about one or more IoTHubs. So depending on rules you have set, the DPS routes the device to the right IoT Hub. Imagine a device on a ship going around the world and always connects to the nearest IoTHub for the best connection (lowest lag). You can also program rules yourself using Azure Functions.

Note: If you want to make use of a DPS with Azure IoT Edge, only a TPM is supported at this moment (2018Q4).

How to register your IoT Edge device

So we need an IoT Edge device like a Raspberry Pi with a TPM on top of it or an industrial PC like the Advantech Uno 2372G with a TPM built in.

At this point, the IoT Edge documentation get’s a bit fussy. All examples I found until now are referring to the use of a TPM emulator, not an actual TPM.

In this blog, I will show you how you can register using a DPS.

Create your own local Azure IoT Edge dashboard

Earlier this year, when Azure IoT Edge was still in Public Preview, I wrote a couple of blogs about Visualizing Azure IoT Edge using local dashboard.

Back then, I had to do some magic with both a C# IoT Edge module, a custom NodeJS docker container, and a Docker network to get it running.

Since then, a lot has changed. Microsoft already released a ton of new features. a And there is still more to come regarding the Azure IoT platform.

But that awkward local dashboard solution was nagging me. A few months ago, Microsoft introduced a NodeJS module as a first-class citizen for IoT Edge modules.

So it was time to pick up the gauntlet and use NodeJS for this awesome local IoT Edge dashboard:

#tldr;  If you like to dig into the code, zip it, clone it, extend it or even make a pull request, I made this project open source. If you only want to use it the easy-going way, pull it from docker eg. ‘svelde/localdashboard:1.0.1-amd64′.

At this moment, only Linux containers are supported. It is tested both on Windows and Ubuntu as host OS.

Interested in this module? Let’s see how you can use it.

How to give a friend full access to Azure Accelerators

Microsoft is providing multiple solutions regarding IoT. If you want to do everything yourself, go for the PaaS services. If you have no programming skills at your disposal, go for IoT central which is a full SaaS solution for IoT.

But if you have skilled programmers in your team and you want to use the full power of Azure IoT, look at the Azure IoT solution accelerators (fka Azure IoT Suites).

There are several accelerators:

At this moment, Microsoft provides five accelerators. One is built by partner Intel (Intel Connected Logistics platform). One is just a simple one for starting to generate lots of telemetry messages (Device Simulation).

The other three are platforms built by Microsoft themselves and these are a great starting point to start building your own IoT platform.

For example, let’s look at the Connected Factory platform. You can create it within fifteen minutes:

Just let a wizard create the resources in your own subscription:

And then you have a running solution.

Do you want to change it? The code and ARM templates of this Connected Factory accelerator can be found on Github.

The same goes for the other accelerators!

For me the most tricky part is something trivial: how do I give a colleague access to these raw diamonds?

Upgrading your Azure IoT Edge device remotely to the latest release

Last week, the Microsoft Ignite event was hosted in Orlando. As with every Microsoft Event, a lot of new stuff is released by the Microsoft product teams.

The same goes for the IoT Edge runtime, the portal and the logic running on IoT Edge.

The IoT Edge runtime bumped up to version 1.0.2 and there are lots of new features introduced like extended offline operation and proxy support.

And there a lot of bugfixes which is always good 🙂

But the documentation considering the update of your gateways is a bit limited.

If I follow the update procedure, I have to update it on the machine itself.

Let’s check out how we update our runtime remotely.

Update: Microsoft released this document with more details after I wrote this blog. Some parts could be out-dated by now.

Prototype an IoT Dashboard with Adafruit IO

Last week I built this great demo with an Azure IoT Edge running on Industrial hardware, reading temperature, humidity, fan activity and led activity. But there was something missing…

I needed a simple dashboard to represent the values which were ingested by my Azure IoTHub and sent to an Azure Function.

Normally I build a basic website myself or I use tooling like PowerBI. It’s not that hard to get something sufficiently running for a demo.

But the last couple of weeks I was looking around for generic, off-the-shelf IoT Dashboards. And I had a couple of questions about their capabilities. What is on the market? What connectivity do they use? How many messages can I Ingest per time window? How do I configure the visual components? Etc.

I have reviewed a number of them and then I was checking out Adafruit IO.

This is what they see about themselves:

“Our simple client libraries work with the most popular devices such as the Adafruit Feather Huzzah, ESP8266, Raspberry Pi, Arduino, and more.”

I was triggered by the ‘more’ part. Does it also work with non-Adafruit devices? Because I know Adafruit from their DIY electronics shop, I was interested in what they are offering. And I was pleasantly surprised.

Let’s take a look at how we can integrate Adafruit IO in a generic demo with industrial hardware.