In this book, in Developing our batch analytics with Airflow section in Chapter 8, Implementing a Custom Industrial IoT Platform, in AWS IoT Analytics section Chapter 10, Implementing a Cloud Industrial IoT Solution with AWS, in Dataflow section in Chapter 11, Implementing a Cloud Industrial IoT Solution with Google Cloud, and in Azure analytics section in Chapter 12, Performing a Practical Industrial IoT Solution with Azure, we discovered how to build analytics based on the cloud or, generally speaking, a centralized infrastructure. We also learned about cold paths and hot paths. From a theoretical point of view, the implementation of analytics should be agnostic with regard to where they are deployed and how we push data. Unfortunately, this is not always the case.

Analytics are strictly coupled with the support they want from the infrastructure. When analytics require a large amount of data to be sampled every millisecond, we might not be able to deploy on the cloud due to the restriction of the bandwidth. If the analytics that we are using require more data than produced in the last few seconds or minutes, a data stream platform may not be the right choice. If the analytics are stateless, we might be able to use serverless technologies. If a user needs to interact with these analytics, they cannot work easily in streaming mode.

In conclusion, the implementation of the analytics depends strictly on how we want to consume data and how we want to interact with the other actors, the device, and the user.