Azure IoT Loves Open Source
azure.microsoft.com
azure.microsoft.com
So just in case anyone is curious here are a bunch of links to just how much we're doing in the open now:
(1) Azure repo collection: https://github.com/Azure (2) Microsoft general repo collection: https://github.com/microsoft (3) .NET Core open source repo: https://github.com/microsoft/dotnet
For all the haters here below, who "actually" blame one developer, who shares his code ( because he posts a project) and he hasn't got another contribution on his new account...
Let me remind you, Microsoft is the #1 contributer on Github now, topping Google and Facebook together. https://news.slashdot.org/story/16/09/15/2255241/microsoft-h...
Please, now let's go back to IoT and ontopic. Too bad i don't know a lot of IoT on Azure, but i have viewed some projects about IoT of Microsoft on Hackster, their project page: https://www.hackster.io/microsoft/products/microsoft-azure
Thanks dluc for sharing, i hope i have topic worthy stuff to contribute next time ;) - one point of discussion would be to link to the project page instead next time - https://github.com/azure/toketi-iothubreact or https://azure.microsoft.com/en-us/blog/developer-s-introduct..., it would probably fuse some less discussion - we like code :p
Azure was their choice, not mine, but by the end of the development I was pretty pleased with the level of security they built into the data acquisition side.
Before the Azure project I worked with some other IoT/Cloud providers and it was a lot shakier. As in, am HTTP GET with the device serial ID in the URL and that's it.
My customer showed me the dashboarding they've already built with BI on the backend, it was even more advanced than the stuff I've seen from the others. Azure is no cakewalk to set up, but it seems to work really well once it is.
My team and I spent a few weeks working with Azure IoT before instigating a switch to an alternative.
The problem with Azure IoT as a platform is that they provide SDK documentation (which is relatively good for languages they like), but not API documentation for the AMQP interface which is their preferred method of connecting devices. Their support for python clients and servers (in particular servers) left me and my team spending circa 10 days digging through AMQP logs and wireshark in ever more desperate attempts to integrate our existing python server with Azure IoT. Eventually it became clear that our only option was to wrap the C library for a really brittle integration and that we could switch to a different platform with far less effort! Both Amazon IoT and Watson IoT have been far smoother to integrate with since they provide both SDK and API docs!
The tough thing about it is that on the surface it looks as though it's really well documented and it's not until you hit your first big problem that you start looking around for API docs to try and debug what exactly it is that's wrong!
Currently really fond of Watson IoT for the decent python SDK's (we've only got 2-3 patches in a fork!) and the really handy device management features which can make rebooting etc much easier from the server side. The MQTT implementation they've worked on is great too.
Good example: https://github.com/BitFunnel (full-text-seach from bing)
Worst-case scenario people would just rewrite parts of the stack to move around it. There's people salivating at this very moment just thinking about rewritting core stuff in Go, Rust, or something else, and making a blog post about it. Imagine if they had an actual motive to do it.
Not sure there is a solution where that isn't the case, but it'd be nice to be wrong.
I believe that in most cases data from smart device should not leave their local network.
Making a product that leaks data, especially from people homes is dangerous for the their users (I do not say owners, as who owns the data is an actual owner in my opinion) and even more dangerous for the society as whole.
> Working on Microsoft Azure iot
and this post is his only contribution.
thanks for reading.