PDU's Core Mission

"All kinds of PostgreSQL data recovery, solved in one tool."

Four Critical Scenarios

PDU is designed to handle the four most critical PostgreSQL disaster scenarios that traditional methods struggle with:

Corrupted Databases

When PostgreSQL cannot start due to data file corruption, PDU extracts data directly from the damaged files—bypassing the database engine entirely.

Use Case:

Disk failure corrupted system catalogs. Database won't start. PITR backup chain is incomplete.

Accidental DELETE/UPDATE

When data is accidentally deleted or updated, PDU recovers the original data by parsing WAL archive files—without needing a full database restore.

Use Case:

Developer ran UPDATE without WHERE clause. 500,000 customer records modified. Need to restore original values.

Data File Loss

When specific data files are lost or corrupted, PDU can work with the remaining files to reconstruct as much data as possible.

Use Case:

Accidental rm command deleted some table files. Need to recover what remains.

UNIQUE

DROP TABLE Without Backup

PDU provides the only publicly documented practical method to recover dropped or truncated tables without backups, using disk fragment scanning.

Use Case:

Critical table dropped. No backup exists. Traditional recovery impossible.

Three Recovery Methods

PDU provides three distinct recovery methods to handle different data sources:

Direct Data File Extraction

Read and parse PostgreSQL data files directly, extracting records even from corrupted files. PDU automatically detects and skips damaged blocks while recovering valid data.

Single table filesEntire database directoriesTOAST data handling

WAL Archive Scanning

Parse WAL (Write-Ahead Log) archive files to extract deleted or updated records. Supports both time-range and transaction-based recovery modes for precise control.

DELETE recoveryUPDATE recoveryTransaction mode

Disk Fragment Scanning

Scan raw disk blocks to find and reconstruct data from dropped or truncated tables. This breakthrough technology makes previously impossible recoveries feasible.

DROP TABLE recoveryTRUNCATE recoveryIndex-based scanning

Flexible Operation Scope

PDU supports recovery at multiple granularity levels, from single tables to entire instances:

Single Table

Recover a specific table by providing the data file and column types

Schema / Database

Export all tables in a schema or database with auto-discovered metadata

Custom Files

Work with specific data files copied from any source location

Output Formats

PDU generates multiple output files to facilitate easy data reimport:

CSV Files

Table data in standard CSV format, ready for import with COPY command

COPY Statements

Ready-to-use COPY commands for each exported table

DDL Statements

CREATE TABLE statements for table structure recreation