Data migration & transformation glossary
Plain-English definitions for the terms that show up in every enterprise data migration RFP.
- AI-Assisted Data Migration
- An approach to moving data between systems where AI models accelerate the most expensive phases — schema discovery, source-to-target mapping, dialect code generation, and reconciliation — while human data architects review and certify the output.
- Portable Ontology
- A vendor-neutral business model (customers, accounts, orders, invoices, etc.) that sits above any specific data warehouse. Because the ontology is portable, the same migration can be re-targeted from one cloud platform to another without rewriting business logic.
- Source-to-Target Mapping
- The specification that defines how each column, table, and transformation in a legacy source system maps to the new target platform. OpenOntos generates these mappings automatically and presents them for human review.
- Bronze / Silver / Gold (Medallion Architecture)
- A layered lakehouse design popularized by Databricks. Bronze = raw ingested data, Silver = cleansed and conformed, Gold = business-ready aggregates and marts. OpenOntos generates code for all three layers in the target dialect.
- Dialect Codegen
- Generating SQL or transformation code tailored to a specific target platform's dialect — Snowflake SQL, Databricks Spark SQL, T-SQL for Microsoft Fabric, BigQuery Standard SQL, etc. — so the output runs natively without a proprietary runtime.
- Data Migration Assessment
- A pre-migration report quantifying scope, effort, risk, and a phased rollout plan. OpenOntos produces the assessment in 1–2 weeks; certified Data Architects review and sign off (human-in-the-loop).
- ETL Modernization
- Replacing legacy ETL platforms (Informatica PowerCenter, IBM DataStage, SSIS, Talend, Pentaho) with modern ELT patterns built on cloud-native compute (dbt, Spark, native warehouse SQL).
- BYO-LLM
- Bring Your Own Large Language Model. OpenOntos lets enterprises plug in OpenAI, Anthropic Claude, AWS Bedrock, Azure OpenAI, Google Gemini, or a private/self-hosted model — so AI provider choice is never a lock-in.
- Human-in-the-Loop (HITL)
- A workflow where AI generates artifacts and a human expert reviews, edits, and approves them. Every OpenOntos migration plan and mapping is HITL-certified by a Data Architect before code is shipped.
- Re-target vs Rewrite
- When a business changes cloud data platforms, a re-target swaps the target layer while preserving the ontology and mappings (hours-to-days). A rewrite starts over (months-to-years). OpenOntos is designed to make every migration a re-target next time.
- Reconciliation
- Row-count, business-key, and aggregate comparison between source and target after migration. OpenOntos reconciles automatically and produces a parity report so you can prove cutover safety.
- Lakehouse
- A data architecture combining the low cost and openness of a data lake with the performance and governance of a data warehouse. Snowflake, Databricks, and Microsoft Fabric are all lakehouse platforms.