fromover

Fromover is a term used in speculative data engineering to denote a pattern of data integration that combines information from multiple disparate sources over time into a single, coherent dataset. The concept emphasizes temporal layering, provenance tracking, and conflict resolution. The name likely derives from the idea of gathering data "from" multiple origins and overlaying (over) them to produce a consolidated view.

In practice, fromover describes a workflow where data streams and batch sources are ingested concurrently, with each data element carrying metadata about its source, timestamp, and quality. A fromover pipeline reconciles changes across sources according to predefined policies, such as source-of-truth designation, versioned overrides, or probabilistic fusion. The approach aims to improve coverage and resilience when individual sources are incomplete or intermittently available.

The concept has appeared in theoretical discussions and experimental demonstrations rather than as a formal industry-standard methodology. It is often contrasted with traditional ETL paradigms and with building data lakes or lakehouses; fromover focuses on overlaying source data in a governed, time-aware manner rather than performing one-shot extractions.

Advantages include better fault tolerance, richer historical context, and improved traceability through provenance metadata. Challenges involve complexity, operator overhead, data quality heterogeneity, privacy concerns, and the need for robust governance.

See also: data fusion, data provenance, data integration, temporal databases.