Menu ▾ ▴

Concept File Structure

Maruf Çetin

Concept: File Structure and Extension Reference

Turkce Dokumantasyon | English Documentation

Category: Core Concepts & Architecture
Subsystem: Physical Storage & File Formats (AmberDB::Base)
Entry Type: File Format & Extension Reference

1. Definition and Overview

AmberDB employs deterministic and purpose-driven file extensions across its physical storage tier. This file architecture is categorized into three distinct classes:

  1. Authoritative Master Data: The single source of truth. If lost, these cannot be derived from elsewhere; they must be backed up.
  2. Derived Secondary Indexes: Inverted indexes and bitsets precomputed from master data. If corrupted or missing, they are reconstructed deterministically via $adb->set_index().
  3. Runtime & Recovery Files: Transaction undo journals, RAM-disk caches, staging buffers, and backup archives.

2. Comprehensive File Extension Reference Table

Extension Classification Reconstructible? Description
.db Authoritative Master Data ❌ NO Berkeley DB (DB_File Hash) primary document table.
.del Authoritative Master Data ❌ NO Soft-deleted records recycle bin (keep_deleted).
.aut Authoritative Master Data ❌ NO Chronological user action and change audit ledger (log_owner).
.cnt Authoritative Master Data ❌ NO High-concurrency atomic view/hit counter store (use_counter).
.unq Authoritative Master Data ❌ NO Bidirectional string-to-foreign-key dictionary and uniqueness index (_${blk}.unq).
.inx Derived Secondary Index YES Primary 8-byte packed binary index containing all active record IDs.
.fld Derived Secondary Index YES Field value $\rightarrow$ IDs inverted exact-match index (match_block).
.src Derived Secondary Index YES Word tokens $\rightarrow$ IDs phonetic full-text search index (search_block).
.fac Derived Secondary Index YES Columnar bitset index for multi-dimensional facet filtering (facet_block).
.srt Derived Secondary Index YES Monotonic pre-sorted binary array of record IDs (sort_block).
.slg Derived Secondary Index YES Bidirectional URL slug map (_0.slg ID $\rightarrow$ Slug, _1.slg Slug $\rightarrow$ ID).
.jinx Derived Secondary Index YES 8-byte packed primary index for cold/junk records (use_junk).
.jfld Derived Secondary Index YES Inverted field match index for cold records.
.jsrc Derived Secondary Index YES Full-text search index for cold records.
.table Schema Definition ❌ NO Table schema configuration file (schema/*.table).
.dbase Schema Definition ❌ NO Database group configuration file (schema/*.dbase).
.amberdb Backup Archive — Compressed, SHA-256 verified portable database archive.
.csv Continuous WAL — Append-only daily audit stream (backup/YYYY/YYYY-MM-DD.csv).
.txn ACID Journal Transient Active transaction undo-journal rollback file (txn/*.txn).
.cache Shared Memory YES RAM-disk shared memory cache file (cache/*.db).
.tmp Disk Buffer Transient Transient staging buffer file under buffer/ (buffer_write).
.lock Process Mutex Transient Operating system flock synchronization mutex lock file.

3. Storage Efficiency and Backup Optimization

AmberDB's native .amberdb archiving utility (AmberDB::Tools->dump) deliberately excludes derived secondary indexes (.inx, .fld, .src, .fac, .srt, .slg) from backup archives.

Consequently, a 10 GB database compresses down to 500 MB - 1 GB, containing only authoritative tables (.db, .del, .aut, .cnt, .unq) and schemas (.table). Upon restoring (restore), the engine reconstructs all secondary indexes automatically with zero data loss.

MongoDB Logo MongoDB