Glossary

Every acronym and platform-specific term used across this documentation, in one place. Terms are defined here once; the pages that use them link back to this list rather than redefining them.

Components

Term	Meaning
PDQ	The Simplitics data platform as a whole — the unified architecture that the components below make up.
AME — Active Metadata Engine	The central metadata repository and API hub. Stores all logic, every operation that has run and every planned operation. Every other component reads its configuration from here.
INGEST — Data Importer	The orchestrated extraction agent. One container per source connection, running on a schedule, landing data unchanged.
DLS — Data Lake Service	The event-driven service that archives, standardises, profiles and publishes delivered files.
DWA — Data Warehouse Automation	The metadata-driven engine that generates SQL and orchestrates loading into the integrated model.
QPI — Quality Performance Indicator	A named data quality check that runs SQL against the warehouse and returns a pass/fail verdict with a message and a measured value.
DataOps Console	The operator-facing web application for monitoring daily execution and resolving failures. Read-and-intervene, not a configuration tool.
Config UI	The developer interface for managing metadata definitions, backed by Git integration and CI/CD deployment. Also written DevUI, Developer UI and User Interface Toolkit in older material and in some release notes; this documentation says Config UI throughout.
Caddy	The reverse proxy in front of the platform, handling TLS termination and authenticated access.

Data processing zones

The zones data passes through, in order. Zone names are configurable in Settings, so an installation may display its own labels.

Zone	Purpose
Landing	The entry point, where delivered data first arrives. A transient buffer — files are processed onward, not retained.
Raw Archive	Immutable, permanent record of every file ingested, organised and tagged for audit. Write-once; it is the recovery path.
Trusted	Processed and published in a standardised, easily consumable format. Deviations against the data contract are recorded here, not blocked.
Profile	Analysis of what actually arrived — discovering new and missing data points without interrupting the load.
Published	The selected table format for the agreed source delivery, providing SQL-based access. Also called Bronze.
Base	The ensemble integration step between Published and Core. Also called the Ensemble Model.
Core	The integrated core business concept model, using ensemble data modelling methodology. Also called Integrated or Silver.
Business	Use-case-driven business data sets, carrying domain-specific definitions and rules. Also called DM, Data Mart or Gold — the console labels it Data Mart in Data Lineage.

Modelling

Term	Meaning
Source Export	A configured extraction job from a source system — a database query, an API call or a file fetch.
Connection	The technical link to a source system: credentials, URLs, paths.
Sourcefile	The definition of an ingested file's structure — fields, types, keys and governance flags. Also the data contract.
Data contract	The agreed structure and typing of a source delivery, held in the sourcefile's fileStructure. Drift against it is what Contract Validation reports.
Model	A target warehouse table with typed attributes, business keys and relationships.
Mapping	A link between a sourcefile field and a model attribute, optionally carrying a transformation.
Mapping group	A logical unit of work: a collection of mappings from one source file to one or more target models, sharing a key mapping definition, filters and sort orders.
Business key (BK)	The attribute or attributes that uniquely identify a record in a model. Composite keys use businessKeyOrder to fix the position of each part.
Ensemble / Base model	The agile, highly normalised integration layer — every attribute and relationship in its own physical table, fully historised. Answers "what did we know, and when?"
CORE / Enterprise model	The consolidated physical copy of the ensemble: one row per business key, the latest version. Answers "what is true now?" quickly.
Loading pattern	How a model is populated: none, full, dayspan, incremental or transaction. none leaves the data as a consolidated view without materialising a physical table, and is the default.
Load Group	A grouping that falls out of the mappings — all attribute loads, say. It has no machine function; it exists to make the order and shape of the loading legible, and to let you filter the console by it.
Load Step	A single action within a Load Group, such as one action in Base. Like Load Group, a result of the mappings rather than something configured.
Additional Task	Custom work chained onto the automatic flow, before or after the generated loads.

Processing and governance

Term	Meaning
CDC — Change Data Capture	Exporting only records that are new, changed or deleted since the last run. In INGEST the agent keeps a local baseline file, so the comparison survives a restart but not the container being destroyed or removed.
Data profiling	Automated checks and validation of a delivery against its data contract, run as part of the flow when enableProfiler is active.
_HID — Hierarchical ID	A dot-delimited path such as 1.2.4.1 recording an element's exact position in a nested source document. Makes nested rows traceable and acts as a stable join key when arrays are split into separate tables.
Granular checksum	A checksum calculated at every object level rather than only at the document root, so an edit to one child record does not mark the whole file as changed.
__fileKey	The identifier stamped on a delivery, used to trace loaded rows back to the file they came from.
SCD Type 2	Slowly Changing Dimension pattern: each change creates a new version rather than overwriting, with a validFrom field marking validity.
Idempotent	Safe to re-run. DWA loading tasks are idempotent — rerunning one cannot duplicate or corrupt data — which is why restarting a failed load is the normal first response rather than a last resort.
Active metadata	Metadata that does not merely describe the system but drives it — lineage, versions and compliance fall out of execution rather than being maintained separately.
Design as Documentation	The principle that the design is the documentation, and that documentation generates the executable code — so the two cannot drift apart.
Data Modifier	The DLS component that transforms JSON, CSV and XML into one common structure, so a delivery can be read through SQL regardless of the format it arrived in.
Publisher	The DLS step that writes the trusted data into the published tables. It is part of the pipeline and cannot be switched off.
KEEPDATAINTRUSTEDFORDAYS	How many days back Published can be rerun from without reprocessing from Raw. Default 3. It is a performance setting, not a retention policy that deletes Trusted.
authp	The internal authorisation group namespace — authp/admin and authp/user. At installation each group is mapped to the GUID of an Entra group.

Operations

Term	Meaning
The daily pass	The proactive routine run every operational day, verifying every delivery and processing step before the daily deadline. See The daily pass.
On file arrival	The event-driven schedule type: the load runs when a delivery arrives rather than on a clock. Written On event-basis, ad-hoc and File-triggered elsewhere; this documentation says On file arrival.
Checklist view	A console view comparing configured against executed. The only thing that catches work which never started — a failure count reads zero for a task that was never scheduled.
Silent failure	A task that completes successfully but returns zero rows. Green everywhere except the export row/file counts. See Silent failure modes.
Expected	An assumption made by an operator or developer, meant as a pointer rather than a verdict. On INGEST Tasks it comes from the configured schedule; on DLS Trace from the sourcefile's expected files per day, prorated across the window.

See also

• Architecture — how the layers fit together
• Platform guide — how the layers connect, and the way into the per-layer detail
• FAQ — the questions that come up most often