Tantor Special Edition DBMS

Increased performance as compared to PostgreSQL

Transparent data encryption

The Transparent Data Encryption technology, implemented as an extension, ensures reliable protection of information at rest. It allows selective encryption of individual tables, enabling flexible security levels for different data categories. Transparent Data Encryption supports various encryption algorithms, including GOST, which comply with modern security standards

Compression in libpq
Learn more ↗

The libpq_compression parameter has been added to control available traffic compression methods between client and server

Optimized SLRU cache behavior

• Optimization of the pglz compression algorithm
• Compression speed improved by ~40% due to the elimination of potentially redundant operations

Increased streaming data processing speed for ARM64 architecture processors

Performance improvements in operations with json/text data

64-bit transaction counter (XID)

PostgreSQL has a limitation (N = 2³²) on the number of transaction IDs (XID), after which a freeze operation is required. With a 64-bit XID, transaction counter overflow becomes virtually impossible

Extended use of SIMD instructions

SIMD technology enables multiple data operations in a single CPU cycle. Enhancements that will only appear in PostgreSQL 18 are already available in Tantor Postgres 17.5.0

WAL prefetching

Improved performance through preloading of WAL (Write-Ahead Log) blocks

More diagnostic information

• Internal DBMS error diagnostics added (parameter backtrace_on_internal_error)
• Server log entries now provide more details at startup
• The pg_controldata utility includes additional metadata about the release and build, simplifying troubleshooting

Configurable SLRU buffer sizes

Improved SLRU (simple least-recently-used) cache performance through buffer structure optimization

Features beyond conventional PostgreSQL

Tantor PipelineDB
Learn more ↗

• Stream processing and optimized time-series storage
• Real-time data processing using only SQL
• In-database ETL — continuous data transformation via SQL queries
• Data volume reduction — only the output of continuous queries is stored and incrementally updated as new data arrives

Autonomous transactions support

• Support for autonomous transactions using the PRAGMA AUTONOMOUS_TRANSACTION syntax ↗
• A background worker pool is used to improve performance

Transparent anonymization

The transparent anonymization mechanism allows sensitive data in the database to be masked or replaced. After masking rules are defined, anonymized data can be accessed using dynamic masking, which hides information only from untrusted users

Performance improvements

The pg_ivm extension has been added, implementing incremental materialized views. This extension enables updates to materialized views as source table data changes

Additional contrib modules

Over 20 additional modules (extensions) have been added to the contrib package

Functionality for generative AI

High-performance columnar data storage optimized for append-only data on modern hardware platforms — vectorized processing and in-memory computation:

• for recommendation systems and content personalization
• for semantic search and natural language processing
• for computer vision and pattern recognition
• for integration with modern machine learning models

Columnar data storage

High-performance columnar storage optimized for append-only data on modern hardware platforms: vectorized processing and in-memory computation

Background data archiving

pg_archive extension — automatic archiving of data from partitioned tables. Configurable auto-archiving policies for outdated data in the background. Combined with the pg_throttle extension, it forms a powerful tool for information lifecycle management (ILM). Columnar access method is supported

Transaction_timeout parameter

The timeout applies to both explicit transactions (started with the BEGIN command) and implicit transactions corresponding to single SQL statements

Tantor Platform ↗

An integrated graphical management and administration tool built into the DBMS:

• Includes a module showing the current wait event of any specific process
• Displays statistics for detected predicates in WHERE clauses and JOIN conditions
• Includes the pg_trace extension for deep SQL query analysis and profiling. Its simple interface enables technical specialists to quickly identify and resolve performance issues

Precision statistics collection

Allows users to find an optimal balance between the accuracy of collected statistics and the time required to gather it

Automatic query plan analysis

Provides insight into query performance without manual intervention, helping to proactively detect inefficiencies

pg_archive extension

Enables automated archiving of data from partitioned tables with customizable background archiving policies. Combined with the pg_throttle extension, it forms a powerful toolkit for information lifecycle management (ILM). Supports columnar data access

Unified GUI suite to manage corporate Postgres databases, and other essential tools

pg_throttle extension

Provides the ability to limit the execution speed of queries. Especially useful in analytical systems where restricting server resource consumption by heavy queries can significantly reduce their negative impact. In the new version, cgroups support has been added. This allows assigning a specific user's backend process to a pre-created control group, providing a workload isolation mechanism to improve system stability.

Execution plan statistics module

A built-in module that tracks execution plans for all SQL statements processed by Tantor DBMS, enabling deeper performance analysis and query optimization.

Hypothetical (virtual) indexes

Support for virtual indexes that allow evaluating the potential performance impact of index creation without actually modifying the database schema

File system–level performance analysis

Integration of the pg_stat_kcache extension, which captures real physical read and write operations, helping to identify performance bottlenecks that are undetectable by conventional tools

Full and incremental backup

Fast backup without downtime, including differential and incremental methods

WAL prefetching

Ability to restore pages, as well as visibility maps and free space maps, in case of WAL corruption

Transaction log integrity control

Automatic detection and recovery from the transaction log buffer

Compatibility and integration with RuBackup

Tested for compatibility and integrated with the RuBackup backup system by Astra Group, with synchronized roadmaps

Transparent data encryption

The transparent data encryption (TDE) technology, implemented as an extension, provides reliable protection of information stored on disk. It allows selective encryption of individual tables, enabling flexible security levels for different data categories. TDE supports various encryption algorithms that comply with modern security standards.

pg_anon extension

Search and anonymization of sensitive data within the database

Credential validation

Validation of credentials during user creation, password changes, and renaming — defining a set of rules for user management and password policies

Enhanced auditing

Advanced auditing within the DBMS, with detailed logging of session and/or object activities

Modern OAuth 2.0 authentication protocol

Integration of the OAuth 2.0 standard significantly enhances security in user authorization. This modern protocol allows applications to access data without transmitting passwords, greatly reducing risks of unauthorized access and credential compromise

Dedicated application for DBMS security auditing

The innovative pg_sec_check application included in the DBMS distribution enables administrators to regularly perform comprehensive vulnerability audits. The tool automatically detects potential threats and generates detailed reports with specific recommendations to improve DBMS security

Transparent data masking

Real-time transparent data masking within the DBMS ensures sensitive information is hidden from untrusted users on the fly