Reverse Engineering an Undocumented Database: Map Tables Before You Migrate
You inherited a critical database with no ERD, no data dictionary, and staff who left. Tribal knowledge is the only documentation, and every change or migration estimate is a guess until someone maps what is actually there.
We reverse engineer the schema so you can maintain or migrate with confidence.
Sound Familiar?
These are the exact issues CTOs and IT directors face with an undocumented database:
- You inherited a production database with no ERD, no data dictionary, and no column comments
- The engineer who understood the schema left; stored procedures and business rules live only in tribal knowledge
- Foreign keys were dropped for performance; nobody can say which tables actually relate
- Every schema change is trial-and-error because cryptic table names hide what they really store
- Migration and modernisation estimates keep slipping because discovery takes months before real work starts
Manual schema discovery commonly burns 3–6 months before migration work begins. Up to 80% of IT outages are tied to human error in environments where tribal knowledge and undocumented dependencies are the norm. Waiting until cutover to learn the schema is how programmes overrun.
What Database Reverse Engineering Actually Delivers
Opaque schema in → tables mapped → relationships recovered → documentation your team owns.
Connect & Inventory
Read-only access or a restored copy; every object catalogued with volume and shape
Recover Relationships
Infer keys and joins from data and usage when foreign keys were never declared
Map Business Logic
Procedures, triggers, and status codes explained in plain language
ERD & Dictionary
Living documentation so maintain, migrate, or API-wrap without tribal knowledge
Everything You Need for Database Documentation You Can Trust
Schema Inventory
Every table, view, column, index, and constraint catalogued with row counts and naming patterns, even when declared keys are missing.
Relationship Discovery
We recover foreign-key relationships from data profiling and join patterns when constraints were dropped years ago for speed.
Stored Procedure Mapping
Triggers, functions, and procedures get dependency graphs so business rules stop living only in one person's head.
Business Rule Extraction
Cryptic abbreviations become plain-language definitions. Status codes, flags, and calculated fields are documented for the next team.
ERD & Data Dictionary
You leave with entity-relationship diagrams and a searchable data dictionary your ops and engineering teams can actually maintain.
Migration-Ready Artefacts
Outputs feed cutover planning: orphan tables, dead procedures, circular dependencies, and risk notes for the modernisation roadmap.
Platforms We've Reverse Engineered
From 5 Months of Guesswork to a Mapped Schema in 6 Weeks
How a mid-market manufacturer stopped depending on one departing DBA and de-risked a cloud migration with full database documentation.
The Undocumented Estate
- 420+ tables on SQL Server with almost no declared foreign keys
- Cryptic names and zero column comments; new hires needed weeks to touch anything
- Business rules buried in 180 stored procedures only one contractor understood
- Cloud migration quote assumed 4 months of discovery before build could start
- Every production change was trial-and-error against live data
The Mapped Schema
- Full ERD and data dictionary covering tables, recovered relationships, and orphans
- Procedure dependency graph with plain-language rule summaries
- Dead-code list flagged 40 unused objects before they polluted the target design
- Migration build started in week 7 with signed-off scope
- Ops and engineering share one source of truth instead of Slack archaeology
Before vs After Database Documentation
How It Works
From first conversation to handed-over ERDs and a data dictionary in 3–8 weeks.
Tell Us What You Inherited
Platform, approximate table count, whether keys exist, and what you need next: maintain, document, or migrate.
Free Scoping Call
30-minute call to size discovery effort, access method, and which artefacts matter most for your risk profile.
Reverse Engineer & Document
We profile the schema, recover relationships, map procedures, and produce ERDs plus a living data dictionary.
Hand Off & De-Risk
Your team owns the documentation. Optional next step: migration plan or API layer built on what we mapped.
Frequently Asked Questions
What is database reverse engineering when there is no documentation?
It is systematic schema archaeology: inventorying tables and columns, recovering relationships that were never declared, mapping stored procedures and triggers, and writing the data dictionary and ERD that should have existed. The goal is that maintenance and migration no longer depend on tribal knowledge.
How long does reverse engineering an undocumented database take?
Manual discovery for enterprise estates often runs 3–6 months before migration work can start. A focused reverse-engineering engagement for a mid-size production schema typically delivers usable ERDs and a data dictionary in 3–8 weeks, depending on table count, missing keys, and procedure complexity.
Which database platforms can you reverse engineer?
We regularly work with SQL Server, Oracle, MySQL, PostgreSQL, and ageing desktop estates such as Access, FoxPro, Sybase, and Informix. If you can grant read-only access (or a restored backup), we can profile it.
Will reverse engineering disrupt our live production system?
No. We work from read-only connections or restored copies. Profiling and sampling are scheduled to avoid peak load. Your production workload stays online while we map the schema.
How does this help before a migration or modernisation?
Industry figures put legacy modernisation failure near 70% by year two when teams rewrite what they do not understand, and Oracle reports that over 80% of data migrations overrun, underdeliver, or fail. Mapping tables, relationships, and business rules before you commit budget turns discovery from a mid-project surprise into a planned workstream.
How much does undocumented database reverse engineering cost?
Focused schema mapping and documentation for a mid-size estate typically starts from around R45,000. Large multi-schema estates with heavy stored-procedure logic usually range from R80,000 to R180,000. That is a fraction of the R198,000–R790,000 often quoted for fully manual documentation of a few hundred tables, and it protects migration programmes that otherwise burn months on archaeology.
Stop Treating Tribal Knowledge as Database Documentation
If your critical system depends on people who might leave, and your migration plan starts with months of guessing, reverse engineering is cheaper than finding out mid-cutover.
Tell us the platform, roughly how many tables, and whether you need to maintain, document, or migrate. We'll show you what a mapped schema looks like for your estate.