Glossary

A

Aggregation
Combining multiple records into a single record, typically with summary calculations (sum, average, count, etc.).

C

Chain
A sequence of steps that transforms data. Chains are the executable units within a chainset. Each chain has a unique name and contains an ordered list of steps.

Chain Call
A step that invokes another chain, passing records as input. Enables composition and reusability of transformation logic.

Chain Import
The ability for one chainset to import another chainset, making the imported chainset’s chains available for Chain Call steps.

Chainset
A container that holds related chains. Chainsets can be enabled/disabled, assigned to roles and workgroups, and can import other chainsets.

D

Debug Logging
Enhanced logging mode for a step that shows detailed input, output, parameters, and processing information. Useful for troubleshooting and understanding step behavior.

E

ETL
Extract, Transform, Load - a data processing pattern for extracting data from sources, transforming it, and loading it to destinations.

G

Group
An organizational category within a chainset used to organize chains by purpose, domain, or frequency. Each chain can belong to at most one group.

J

Job
An execution of an ETL chain. Jobs are queued and executed sequentially per chainset.

Job Queue
The queue that holds pending ETL jobs. Multiple jobs for the same chainset execute sequentially; jobs for different chainsets can run concurrently.

R

Record
The unit of data flowing through an ETL chain. Records are typically JSON documents with fields and values. Records stream through steps one at a time.

S

Step
An individual transformation operation within a chain. Steps are the building blocks of ETL chains. Over 800 step types are available across 70+ categories.

Step Category
A grouping of related steps (e.g., Array, String, MongoDB, Number, Date). Categories help organize and find steps.

Step Structure Pattern
Defines how a step transforms the record stream: - 1 => 1: One record in, one record out - 1 => N: One record in, multiple records out - N => 1: Multiple records in, one record out
- N => M: Multiple records in, multiple records out - 1 => 0: One record in, no records out (terminal)

Streaming Execution
The execution model where records flow through steps one at a time rather than loading all data into memory. Enables processing of large datasets efficiently.

T

Test Input
Sample data stored with a chain for testing purposes. Test inputs document expected input format and enable repeatable testing of chain behavior.

W

Workgroup
An organizational unit in Ambience that can be associated with chainsets for access control and filtering.