Public docs

SAP migration

Martenweave is an open-source data model registry for SAP migration, MDM, and data governance. This page describes how migration teams can use it to keep model knowledge stable, validated, and traceable while preparing and executing SAP data migrations.

What the problem looks like

SAP migrations—such as Customer, Vendor, and Business Partner—involve hundreds of fields, multiple tables, legacy sources, and shifting business rules. Knowledge often lives in spreadsheets, tickets, validation reports, and individual memory. That makes it hard to answer questions like:

• Which legacy columns map to which SAP fields?
• What does a field mean in a specific SAP context (company code, sales area, purchasing organization)?
• What breaks if we change a mapping rule or a value list?
• Which datasets are missing expected columns, values, or references?

How Martenweave helps

Martenweave stores model knowledge in canonical Markdown and YAML files. Each object has a stable ID, type, status, and references. Deterministic validators check IDs, types, references, and SAP context rules before anything is indexed.

For SAP migration work this means:

• Field-level mapping registry — link legacy source columns to SAP table/field endpoints with explicit Mapping objects.
• Context-aware definitions — attach EntityContext and AttributeUsage records so the same attribute can mean different things in KNVV, KNB1, KNVP, BUT000, LFA1, LFB1, or LFM1.
• Validation before migration — catch broken references, missing required SAP context, and duplicate IDs before they become runtime errors.
• Dataset gap detection — profile CSV and XLSX datasets and compare them against the canonical model to find missing columns, unexpected values, and coverage gaps.
• Lineage and impact analysis — trace from a legacy column through mappings, attributes, and contexts to every downstream SAP field; run impact analysis before changing a rule.
• Human-approved AI proposals — ask AI to draft a PatchProposal from a migration note or ticket; deterministic validators verify it, and a human approves the change.

What it is not

Martenweave does not write data back to SAP. It does not replace an enterprise MDM platform, run workflows, or operate as a chatbot. The local generated viewer is read-only. SAP table and field names are used descriptively for migration context; Martenweave is not an SAP product, partner, or certified solution.

Example scope

A typical SAP Customer / Business Partner migration registry includes:

• DOMAIN-CUSTOMER-BP — the master data domain.
• EntityContext objects for sales area, company code, partner function, and central data.
• Attribute and AttributeUsage objects for fields such as Customer Group, Tax Number, and Partner Function.
• FieldEndpoint objects for SAP targets like FEP-S4-KNVV-KDGRP.
• Mapping objects from legacy source columns to SAP field endpoints.
• ValueList and ValueMapping records for codes that must be translated during migration.
• Issue, Decision, and ChangeRequest objects to capture gaps, agreements, and approved changes.

The examples/customer_bp_model directory in the core repository shows a full canonical slice.

Getting started

1. Install martenweave-core with pip install martenweave-core.
2. Run martenweave init ./my-model to scaffold a repository.
3. Add your model objects as Markdown + YAML files under model/.
4. Run martenweave validate and martenweave build-index --jsonl.
5. Profile a dataset with martenweave profile-dataset and compare it to the model.
6. Review gaps, lineage, and impact reports before building migration routines.

See Quickstart and How it works for commands and the full workflow.