Prepare
- Upload Software AG Adabas files
Guide
Adabas is a proprietary database management system developed by Software AG with hierarchical and multi-valued field support. It has been widely used in enterprise and mainframe environments.
Why Agent Required
Adabas requires the DataMeans Agent because it runs on mainframe systems that cannot directly upload files to the web. The Agent connects securely to your Adabas environment for extraction.
What You Can Extract
- ASSO and DATA datafiles
- Index files
- Hierarchical data structures
- Multi-valued fields
- Compressed records
What You Get Out
DataMeans extracts your data into multiple modern formats:
| Output | Description |
|---|---|
csv/{TableName}.csv | One CSV file per table with all row data |
xlsx/{TableName}.xlsx | Excel workbook per table |
xls/{TableName}.xls | Legacy Excel format per table |
json/{TableName}.json | JSON array of records per table |
json/{TableName}.jsonl | Newline-delimited JSON (streaming-friendly) |
postgres.sql | PostgreSQL CREATE TABLE + INSERT statements |
schema/schema-graph.json | Relationship graph for visualization |
schema/er-model.json | ER model for diagram tools |
report.json | Structured extraction report |
report.md | Human-readable extraction summary |
How It Works
Agent availability: The DataMeans Agent is currently in development and not yet available for download. The steps below describe the planned workflow. To get notified when Adabas support launches, reach out via our About page.
- Download and install the DataMeans Agent on a system with Adabas access
- Configure connection to your Adabas environment
- Select schemas and tables to extract
- The Agent extracts data and uploads securely to DataMeans
- Review and download your converted data
Planned Support
- Adabas versions 4.x through 8.x
- Record structure extraction
- Multi-valued field handling and normalization
- Hierarchical relationship mapping
- Compression algorithm handling
Known Limitations
- Requires DataMeans Agent installation
- Network access between Agent and Adabas required
- Very large databases may require scheduled extraction
Last updated: January 2026
Technical reference
Overview
Adabas (a contraction of "adaptable database system") is a high-performance transactional database management system developed by Software AG for mainframe and enterprise computing environments. It employs an inverted list architecture for rapid data access and supports complex data structures including multiple-value fields and periodic groups. Unlike traditional relational databases, Adabas organizes data in files with records that can contain repeating groups, optimized for OLTP workloads on IBM mainframes and other enterprise platforms.
History and Background
- 1969: Software AG founded in Darmstadt, Germany.
- 1971: Adabas first released, running on IBM mainframe systems under DOS/360, OS/MFT, or OS/MVT.
- 1979: Software AG introduces Natural, the 4GL development language closely associated with Adabas.
- 2016: Software AG announces its Adabas & Natural 2050 agenda, committing to support the products through the year 2050 and beyond.
- 2017: Adabas 8.4 for mainframes released (version 8.4.1, May 2017).
- 2019: Adabas 8.5 for mainframes released (version 8.5.1, October 2019).
- 2021: Adabas 8.5.4 for mainframes released (October 2021).
- 2022: Adabas 7.0 for Linux, UNIX, and Windows released (October 2022).
- 2024: Adabas 8.6 for z/OS released (version 8.6.1, October 2024).
- 2026: Adabas 7.2 for Linux and Cloud released (version 7.2.0, April 2026), replacing version 7.1.1.
File Format Specifications
Adabas stores each database in a set of physical container data sets, separating data storage from index structures.
Container Data Sets:
- ASSO - Associator (inverted lists, field definitions, and address converter)
- DATA - Data Storage (compressed data records)
- WORK - Work data set (protection data and intermediate search results)
- PLOG - Protection log (before- and after-images of changed records)
- CLOG - Command log (information from the control block of each command)
- ALOG - Audit log container (DD name DDALOGRx), recording audit events for the Adabas Auditing add-on
File Structure:
- Associator: Contains inverted lists, file control blocks, the free space table, and the Field Definition Table of each file
- Data Storage: Divided into blocks, each identified by a 3- or 4-byte relative Adabas block number (RABN)
- Record Structure: Each stored record carries its ISN and compressed field values (default compression and null suppression)
- Multiple-Value Fields: Stored within the record, preceded by a binary occurrence counter
- Periodic Groups: Consecutive fields in the FDT that repeat together within a record
- Spanned Records: From Adabas 8, a logical record can occupy up to five physical records (one primary, up to four secondary)
- Physical Extents: The Associator and Data Storage can each be allocated on up to 99 separate data sets
- Log Limits: Each CLOG, PLOG, and RLOG data set is limited to 16,777,215 (x'FFFFFF') blocks
Key Components:
- FDT (Field Definition Table): Defines each file's fields, formats, lengths, and options
- ISN (Internal Sequence Number): 4-byte logical identifier that uniquely identifies each record and never changes
- Address Converter: Maps each ISN to the RABN of the Data Storage block holding the record
- NI (Normal Index): Lowest level of an inverted list, with one entry per descriptor value listing the ISNs of matching records
- UI (Upper Index): Higher inverted-list levels, each managing the next lower index level
- LOB File: Separate file, tightly associated with its base file, in which values of large object (
LB) fields are stored - Expanded File: Logical file of up to 128 component files with identical FDTs and non-overlapping logical ISN ranges
Data Types and Structures
| Type | Description | Storage |
|---|---|---|
| A (Alphanumeric) | Text strings, left-justified | Variable length, up to 253 bytes |
| B (Binary) | Unsigned binary data, right-justified | 1-126 bytes |
| F (Fixed point) | Signed two's-complement integers | 2, 4, or 8 bytes |
| G (Floating point) | Signed normalized floating-point | 4 or 8 bytes |
| P (Packed decimal) | Signed packed-decimal numbers | 1-15 bytes |
| U (Unpacked decimal) | Signed zoned-decimal numbers | 1-29 bytes |
| W (Wide character) | Wide-character text, Unicode by default | Variable length, up to 253 bytes (encoding-dependent) |
A with LA option | Long alphanumeric | Up to 16,381 bytes |
A with LB option | Large object (LOB), stored in a paired LOB file | Theoretical maximum just short of 2 GB |
DT option on B, F, P, or U | Date-time edit mask (DATE, TIME, DATETIME, TIMESTAMP, NATDATE, NATTIME, UNIXTIME, XTIMESTAMP); no separate format code | Length requirement varies by mask; applied in format/search buffers |
Data Model:
- Files contain records with fields defined in the FDT
- Multiple-value (MU) fields allow up to 191 occurrences per record by default, extendable to about 65,534
- Periodic (PE) groups repeat a set of consecutive fields within a record
- Descriptors enable indexed access; superdescriptors combine all or parts of 2-20 fields
- Unique descriptors require each value to occur in no more than one record
- A component file with 3-byte ISNs holds up to 16,777,215 records; 4-byte ISNs raise the limit to 4,294,967,294
- Relationships between files are implemented through descriptor-based file coupling
Version Differences
| Version | Year | Key Changes | Compatibility |
|---|---|---|---|
| Initial release | 1971 | Inverted-list DBMS | IBM mainframe systems |
| 8.4 | 2017 | INDEXUPDATE and UPDATECONTROL ADARUN parameters; utility time-zone support; AUDITLOG | 8.4 ADALNK link routines can call Adabas 8.3 databases |
| 8.5 | 2019 | ADATCP via Adabas Directory Server; ICU-based encoding support; Adabas Auditing (ALOG container); 8-byte STCK timestamps added to Work and PLOG protection records | z/OS, z/VSE, BS2000 |
| 8.5.3 | 2020 | Adabas Auditing for z/OS and Adabas Encryption for z/OS introduced as separate add-on products; ICS library updated to version 3.1.1 | z/OS; databases with UES enabled must add ICS311.Lxxx to JCL |
| 7.0 (open systems) | 2022 | Current Linux/UNIX/Windows line, versioned separately from mainframe releases | Linux, UNIX, Microsoft Windows |
| 8.6 | 2024 | 64-bit buffer pool above the 2 GB bar (multi-gigabyte LBP); dynamic allocation of SORT and TEMP data sets; expanded SYNS-5B checkpoint records | z/OS (8.6.1, October 2024); TEMPSIZE/SORTSIZE utility parameters optional |
| 7.2 (Linux and Cloud) | 2026 | WORK containers encrypted when database encryption is active; Azure Key Management System support; XI option for unique descriptors in periodic groups | Replaces 7.1.1; ADACVT converts databases in place to and from earlier versions |
Compatibility Notes:
- Adabas 7 introduced Universal Encoding Support (UES) and the wide-character (W) field format, with Unicode as the default wide-format encoding
- Adabas 8 introduced spanned records and large object (LB) fields
- The nucleus and utilities used with a database must match its version; the ADACNV utility converts a database to a higher or lower version level
- Adabas link routines (ADALNK) can be used across versions
- On Linux, UNIX, and Windows, ADACVT converts a database in place between versions; alternatively, an ADABCK backup taken with one version can be restored by the ADABCK of another
- ADATCP (direct TCP/IP access) has been available on the open-systems line since version 6.7.1
- Open-systems versions 6.7.1 and below do not support encryption; an encrypted database must first be restored with ENCRYPTION=NO before it can be used in such a version
Technical References
- Software AG Adabas for Mainframes Documentation
- Wikipedia: Adabas
- Adabas Design (Concepts and Facilities)
- Adabas Field Definition Statements
- Software AG Adabas & Natural Product Overview
To learn how to use this format with DataMeans, see the User Guide.