Design and Implementation of an Operational Backend System Using ETL, Metadata, and Business Object Model

The article describes the evolution of an operational backend system for a video platform, covering three generations: Apollo configuration center (1.0), a custom-developed backend (1.5), and the Eight system (2.0).

It outlines the operational requirements: cross‑business integration, low‑cost development, and good interaction experience.

The technical solution adopts a metadata + ETL model to unify heterogeneous data sources, a business‑object model as the smallest independent unit for read/write operations, and a componentized UI layer that dynamically binds data to frontend components.

Metadata modeling is illustrated with HTTP and ElasticSearch examples, showing source description, input/output attributes, and parsers.

The business‑object model combines fields from multiple sources, enabling flexible assembly of operational features while keeping the platform decoupled from specific business logic.

Componentized UI encapsulates common web controls (select, input, date picker) and binds them to metadata‑driven attribute types, reducing repetitive page work.

The overall architecture follows a micro‑kernel, plugin‑based design using patterns such as adapter, factory, template, responsibility chain, and SDK for low‑overhead data access.

Benefits include platform‑business decoupling, rapid configuration of complex needs, data safety via source/attribute constraints, and dynamically generated interactive pages.

Future work focuses on expanding the UI library, supporting more heterogeneous data sources, adding deep‑operation features (workflow, export, task allocation), and opening the platform as a PaaS or open‑source product.