Case study · Automotive retail group, southern Brazil
20 years of memory recovered from discontinued systems.
A single cloud database and tax, fiscal, accounting and payroll assessments in days, not months.
In numbers
- records recovered from two discontinued systems
- ~230 M
- unique electronic tax documents
- ~1 M
- files processed
- ~8.9 M
- discrepancies in validation against the source
- 0
- of history
- 20+ yrs
- of immutable retention
- 10 yrs
The client
Two decades of operations, scattered.
An automotive retail group in southern Brazil, with 5 companies and more than 10 establishments, whose operations had ended while significant legal and tax decisions were still pending.
Twenty years of memory sat in systems nobody operated anymore: an HR and payroll system on an Oracle database, an ERP on a Progress database, hundreds of thousands of electronic tax documents, the accounting department's PC with backups of even older machines, the tax department's mailboxes and thousands of spreadsheets, documents, PDFs and notes.
The challenge
Three lawsuits and a tax recovery depended on that data.
- Discontinued systems, with no vendor support and no team that knew how to operate them.
- Servers out of operation for more than 5 years, with hardware at its limit: the application server had already lost its disk; the database server was raising memory and cooling alerts.
- Old database versions and legacy code pages, in which accented characters “break” during conversion.
- Lawyers and accountants needing amounts by period, by establishment and by pay item.
- Without the database, the only path was to rebuild everything from paper and loose reports: months of manual work, with a high risk of error and contestable evidence.
The CALMO solution
From network scan to Data Center.
Diagnostic and scan
Read-only network scan; every device classified and every source identified — including a forgotten database inside one of the servers.
Collection under chain of custody
Native exports of the Oracle and Progress databases, packaging of tax documents and of the accounting PC, with SHA-256 computed at source and file-by-file manifests.
Recovery without the vendor
Oracle restored with code-page correction, Progress migrated from its data dictionary and an in-house parser for NF-e and CT-e XML.
Single cloud database
Managed, encrypted, private database, one schema per source system and full-text search: name, plate, chassis or notes of any document.
Cross-validation
Count of every table and every file against the source manifest — zero discrepancies; payroll reconciled with the general ledger and bank payment dates.
Pleading-driven assessment
Parameters extracted from the court decision, calculation guideline validated with the lawyers, workpapers with three scenarios and a technical report.
Retention and Data Center
Everything collected kept in immutable retention for 10 years; the database remains available for new queries by CNPJ and period, including in natural language.
Results
Unqualified evidence, in days.
- Tax assessment delivered in days from the first collection — work that would take months by hand.
- Technical report without qualifications: the database proved the actual payment dates and the absence of installment plans.
- ~1 million unique tax documents — inbound and outbound NF-e and CT-e covering more than 12 years — searchable by CNPJ, period or any information in the documents.
- ~250 thousand files from the accounting PC preserved: e-mail, spreadsheets, documents, local databases and tax software.
- Recurring queries through the Data Center, in natural language, with immediate answers.
| Indicator | Value |
|---|---|
| Group companies | 5 |
| Establishments | more than 10 |
| Discontinued systems recovered | 2 (HR/payroll and ERP) |
| Tables recovered | ~4,700 |
| Records | ~230 million |
| Unique tax documents | ~1 million |
| Files processed | ~8.9 million |
| Discrepancies in validation | 0 |
| Period covered | more than 20 years |
| From collection to assessment | days |
| Immutable retention | 10 years |
Case published with anonymized data and rounded figures.
Does your client have a similar story?
Old systems, idle servers and a demand with a deadline. Start with the diagnostic.