AWS' IoT Analytics product has been released in its Sydney region, allowing uses to run analytics on large volumes of data from internet of things (IoT)-enabled devices.
The service collects specific data, processes it through mathematical transforms, enhances it through device-specific metadata – such as device type and location – and then stores it in a time-series data store.
The stored processed data can then be analysed through running ad hoc or scheduled queries via a built-in SQL query engine, with more complex analytics and machine learning inference available with pre-built models for common IoT use cases.
Additionally, users can run their own custom analyses, which utilise an Analytics Compute Unit (ACU) packaged in a container. Each ACU contains four vCPU and 16 GB of memory.
The execution of custom analyses created in Jupyter Notebooks or other tools – with AWS naming MATLAB and Octave as two examples – can also be automated by AWS IoT Analytics to be executed on a user’s own schedule.
The service is available in both a free and paid tier. In the free tier, users have a monthly allowance of 100 MB of data processed, 10 GB of processed data storage, 10 GB of raw data storage and 10 GB of data scanned for query execution.
In the paid tier, data is processed at US$0.24 per GB, processed data is stored at US$0.036 per GB and raw data is stored in Amazon S3 and charged at the standard S3 rate, which is on a sliding monthly scale of US$0.0025 per GB for the first 50 TB, then US$0.0024 per GB for next 450 TB and then US$0.0023 per GB for every GB after.
Meanwhile, standard SQL query execution is charged at US$7.80 per TB, while custom analysis execution is billed in one second increments and charged at US$0.43 per ACU hour.
Additionally, AWS IoT Analytics contains 40 GB of free EBS storage to temporarily store analysis containers during execution, with more available by contacting AWS.