Azure Stream Analytics Review

Our primary use case is mainly to ingest real time data streams into permanent storage places like databases, block storage, etc.

The biggest improvement for us has been that it now takes much less time for us to receive valuable information. Basically, as soon as it appears in our real time data source, in a matter of seconds, it is already in our database.

The value of this solution is the deep integration it provides with other Azure resources which we use a lot. Our whole infrastructure is pretty much based on Azure so ease of integration is a valuable feature. Secondly, the simplicity of the solution is great. You don't need to set up much, you just make a selection, select a destination, and you're off. 

There are some improvements that could be made, first of all in pricing, because right now the pricing is a bit unclear. It's hard to estimate how much of that is a local issue but you can't figure out how prices are calculated or the proprietary part of the cost. Another area that could be improved is that if something does go wrong, it's very hard to investigate what caused it and why. The logging is available but it lacks detail and doesn't provide much information.

I've been using this solution for two years. 

The solution has an acceptable level of stability although, as mentioned, if it does fail, it's pretty difficult to find out the cause. 

It's very easy to scale this solution. We probably have a couple of hundred users and we have developers who deal with maintenance. This is our main tool for real time data streaming. 

The initial setup is quite straightforward. Because of the good integration, you select your real time data, store the destination where you want to write it and you probably don't even need to transform with data. You basically create a mapping descent source. We had a proof of concept in place, so I would say deployment took two working days without having a deployment plan. 

We have a good ROI because we are able to deliver solutions very quickly and customers are happy with that. 

We evaluated and carried out a comparison with Oracle. The results were pretty much the same for both in terms of real time data streams, but were very much tied to their own cloud solutions. If you work with Oracle i'ts probably best to go with Amazon.

My simple advice would be to not scale up initially. Also, if you have questions don't just rely on the official documentation, but use other resources such as a blog by a developer, because sometimes that can be more helpful than documentation provided by the company.

The best advice I can offer would be that if there is a simple solution available, do not try to complicate things. 

I would rate this solution an eight out of 10.