Azure SQL Database Edge is a fascinating development for anyone working with SQL Server, because it promises to extend the range of applications for the product far more than their previous embedded systems or IoT products.

Well, it certainly fascinates me. I've recently been running versions of SQL Server on progressively smaller devices, as a sort of hobby. I've done a few industrial database-driven applications, the first in about 1980, well before the first commercial SQL Database. I'd have loved, then, something more resilient than the KSAM database system that we used.

Although there is nothing much wrong with small-scale SQL databases such as SQL Lite, they aren't so useful for embedding in appliances because they are only single user. For many years I used SolidDB for embedding into real-time applications because it was so robust, but the idea of using SQL Server has a certain attractiveness, because it requires no extra learning. I notice that plenty of folks are using MongoDB in network appliances nowadays, but I reckon that a well-designed relational database will always give the best response.

It was fun, and miraculous, to see Azure SQL Database Edge working in a container on a Raspberry Pi 4, which can fit in your hand. SQL Database Edge is the SQL Server database engine, even with columnstore and in-memory OLTP, but running on 64-bit ARM devices. It runs on ARM and Intel architecture, without analytical engines or in-memory capabilities, but it is optimized for edge devices. It's just SQL Server, on Linux, on tiny computers that can run anywhere.

It is great for learning the technology, but don't get too overexcited about running a website on it, because licensing for these embedded system products currently only permit it to be installed on the device with an integrated (embedded) software application (or suite of applications) dedicated to a specific use. The idea of running an enterprise application on it is laughable anyway. I can get a second-hand, rack-mounted HP server to host SQL Server for not a great deal more than a Raspberry Pi.

I'm rather wary of running SQL Server in a container, though. So far, I must confess, my experience of running SQL Server containers in Windows or Linux has given me the feeling of being a pioneer. Advice and installation instructions seem to change. Any installation that succeeds gives me a feeling of triumphing over adversity. It all can be made to work but why make it all so difficult? There is a curious retro feel to the experience, as if one were back in the late seventies. I once proudly got a whole range of SQL Server containers working on Windows, only for Docker to withdraw support for the version of the operating system and I lost the lot without warning. It doesn't inspire confidence.