Stream Data Platform - AF uses predefined scenarios to analyze the stream data that has been input into the stream data processing engine and then outputs the results of analysis that you have set up to meet your objectives.
A scenario is defined as a query in a query definition file. The content to be defined in a query is described below.
- Data to be targeted for analysis
Stream data is time-series data that continues without interruption. To analyze steam data, you must specify the range of data to be included.
Using the concept of windows, Stream Data Platform - AF specifies the range of data targeted for analysis by treating stream data as finite data separated based on time stamps or number of tuples.
- Details to be analyzed
The intended analysis is carried out on stream data separated into windows. For example, you can track the change in values for a specific series of data, or combine multiple stream data values into new stream data and analyze it.
You define the details to be analyzed using relational operations such as selection and linkage, operations that use aggregate functions, and the like.
- Formatting the analysis results for output
The results of analysis are output as new stream data.
For example, you can output analyzed stream data when there is an increase or decrease in the analysis results, or output all analysis results at specified time intervals.
A function called stream operation is used to output analyzed data.
You define queries using a query language called CQL, which is similar to SQL.
Details about how to define queries using CQL are explained in 2. Using CQL to Define Queries. Details about how to code CQL are also explained in 3. CQL Basic Items and Data Types and 4. CQL Reference. For details about the query definition file, see the uCosminexus Stream Data Platform - Application Framework Setup and Operation Guide.
You can also define a new query definition file based on a sample file. Details about the sample files are explained in 5. Query Definition Samples.