Recipes¶
Automation recipes are collections of pipelines or processes that leverage tools and inputs to generate analytic insights.
Recipes can be thought of as uni-directional graphs containing a number of discrete nodes, each corresponding to a specific step in a data analytic pipeline. Just like a culinary recipe, recipe graph nodes are always processed sequentially.
Recipe nodes use Kubernetes pods, or containers, to ensure they are distinct mini-processes with processes isolated from other nodes in a graph. These containers are lightweight, and only contain the data, tools, code, libraries, and settings required to run the process for which they are designed.
Recipe structure¶
Every recipe has a set structure, built on standard JSON formats.
- Description: each recipe has a description that appears on the Recipes page and in the recipe's
description.jsonfile. - Variations: each recipe variation is a specific combination of recipe graph, order schedule, and set of variable overrides for processing.
- Recipe graph: each recipe has a graph component that represents the structure of its nodes.
- Recipe variables: variables have their own process order—recipe variables are overridden by kitchen variables, and in turn override recipe variation variables. Recipe variables can refer to the specific filename of a data source or sensitive credentials that are important to hide.
File Editor¶
You can access all underlying recipe files in the UI from the File Editor.
To access the File Editor:
- From the Recipes page: Select a recipe from the list, click Recipe Actions, then select View Files.
- From the Variation page: Click <> Switch to File Editor.
- From the Node Editor: Click <> Switch to File Editor.