TDSQL-A for PostgreSQL

An online real-time data warehouse service featuring high performance, scalability, security, and cost effectiveness

Disclaimer

Please note that this product is not currently in operation and not available for purchase. Any product information set forth below is strictly an overview of our proposed arrangements in relation to this service, which may be subject to change as and when it is made available for purchase on this website. You acknowledge that any information herein is for your reference only and does not constitute or form any part of an offer to sell.

Overview

TDSQL-A for PostgreSQL is Tencent's proprietary distributed analytic database system. It adopts a shared-nothing architecture, offers a proprietary column storage engine with a high compression ratio and a high-performance new-generation vectorized execution engine, and supports hybrid row/column storage. It features the complete atomicity, consistency, isolation, and durability (ACID) capabilities of distributed transactions, compatibility with PostgreSQL and Oracle syntax, and multi-level disaster recovery and security capabilities, providing you with efficient storage and online analytical processing (OLAP) services for massive amounts of data at the gigabytes to petabytes levels.

Benefits
Hybrid Row/Column Storage

TDSQL-A for PostgreSQL offers hybrid row/column storage capabilities with high cost-effectiveness to sustain efficient hybrid row/column computing. Its proprietary column storage supports multiple compression algorithms and levels and features adaptive compression capabilities and a high compression ratio.

Ultra-High Performance Based on Fully Parallel Architecture

TDSQL-A for PostgreSQL adopts a distributed shared-nothing architecture to process inter-node, intra-node, and intra-operator computations in full parallel. Plus, its efficient vectorized execution engine, late materialization technology, and ability to return results for trillions of correlated subqueries within seconds all contribute to the smooth storage and computing of petabytes of data.

High Security and Availability

TDSQL-A for PostgreSQL separates the permissions of security, audit, and database admins, supports transparent data encryption, data masking, forced access control, and comprehensive audit, and features multi-level disaster recovery and high availability.

Smooth Business Migration

TDSQL-A for PostgreSQL is compatible with SQL 2011 standards, PostgreSQL syntax, and Oracle syntax. Plus, it supports stored procedures, views, and triggers, and comes with migration tools for quick migration.

Complete Transaction Capabilities

TDSQL-A for PostgreSQL has complete transaction ACID capabilities to guarantee the global consistency of distributed transactions. Its proprietary patented technologies ensure the consistency and efficiency of the distributed architecture.

Powerful Data Governance Capabilities

TDSQL-A for PostgreSQL is capable of efficient online auto scaling, automated hot/cold data separation, warehousing data in multiple methods, quick interconnection with other databases through FDW-based foreign tables, and quick data migration from TencentDB.

Features

Data Encryption

Support for Column Storage and Multiple Compression Algorithms

Data Masking

Comprehensive Audit

Hot/Cold Data Separation

Multi-Level Disaster Recovery

Data Encryption

TDSQL-A for PostgreSQL provides two data encryption methods:

Encryption on the business side

The business calls the encryption function built in TDSQL-A for PostgreSQL and writes encrypted results to the database. Normally, the encrypted data will be read and decrypted in the application.

Encryption built in TDSQL-A for PostgreSQL

The encryption process is transparent to the business, which has the following advantages:

1. Encryption operations (function calls) are decoupled from the business. The business is only responsible for writing the original data to the database kernel, and the subsequent encryption calculation will be conducted inside the database, which is imperceptible to the business.

2. Encryption algorithms are maintained by the database, and operations such as encryption algorithm selection and key management are all performed by the security admin independently.

Kernel-based encryption calculation supports async encryption to implement data encryption while delivering a stable system throughput. The supported encryption algorithms include AES-128, AES-192, AES-256, and SM4.

Support for Column Storage and Multiple Compression Algorithms

TDSQL-A for PostgreSQL supports column storage. You can define data tables as columnar tables based on your business needs. Generally, we recommend that you set tables with a large width and requirements for a high compression ratio as columnar tables. Columnar tables support a wide variety of compression algorithms, such as delta, zlib, zstd, RLE, and bitpack, which have different compression levels. For more information, please see the corresponding section in the user manual. TDSQL-A for PostgreSQL supports the new-generation columnar vectorized execution engine, which offers high query performance for hybrid row/column storage and query.

Data Masking

TDSQL-A for PostgreSQL supports transparent data masking that can return masked data to unauthorized users in an imperceptible manner.

Comprehensive Audit

TDSQL-A for PostgreSQL supports all-round auditing in multiple dimensions. Bypass detection is used for auditing, which has little impact on database operations. The following types of audits are available:

1. Statement audit: audits a certain type of statement.

2. Object audit: audits the operations of a certain database object.

3. User audit: audits the operations of a certain database user.

4. Fine-Grained audit (FGA): uses expressions as audit conditions and allows you to set actions for when an audit is triggered, such as sending emails or making phone calls.

Hot/Cold Data Separation

The kernel natively supports hot/cold data separation so that the business can provide a unified database view without having to perceive the differences between the underlying storage media.

1. Hot data and cold data are stored in different node groups with different physical server configurations to implement hot/cold data separation and reduce costs.

2. Scheduled backend tasks can automatically migrate data according to the configured hot/cold data rules. This allows the system to implement automated hot/cold data separation and eliminate the need for the business to pay attention to the storage of hot/cold data in the cluster.

This feature is currently available on the Private Cloud Edition but unavailable on the Public Cloud Edition.

Multi-Level Disaster Recovery

TDSQL-A for PostgreSQL ensures the cluster disaster recovery capabilities in multiple dimensions:

Strong Sync Replication

TDSQL-A for PostgreSQL supports strong sync replication. Ensuring that the primary and standby nodes have the identical data is the basis of the entire disaster recovery system. If the primary node fails, the database service can be switched to the standby node without any data loss. The strong sync replication mechanism requires that a success be returned only after the user request is executed and the log is successfully written to the standby node, guaranteeing that the data on the primary and standby nodes are always consistent.

Primary/Standby High Availability

The primary/standby high availability solution of TDSQL-A for PostgreSQL mainly uses multi-replica redundancy in each node group to ensure that there are no or only momentary service interruptions. If the primary node in a group fails and cannot be recovered, a new primary node will be automatically selected from the corresponding standby nodes to continue service provision. Based on primary/standby high availability, TDSQL-A for PostgreSQL supports the following features:

1. Automated failover: if the primary node in the cluster fails, the system will automatically select a new primary node from the corresponding standby nodes, and the failed node will be automatically isolated. The strong sync replication policy ensures complete primary/standby data consistency in case of primary/standby failover, fully meeting the finance-grade requirements for data consistency.

2. Failure recovery: if a standby node loses data due to disk failure, the database admin (DBA) can recover the standby node by building it again or add a standby server to a new physical node to recover the primary/standby relationship and thus ensure the system reliability.

3. Replica switch: each node in the primary/standby architecture (which can contain one primary node and multiple standby nodes) has a complete data replica that can be switched to by the DBA as needed.

4. Do-Not-Switch configuration: it can be set that failover will not be performed during the specified period of time.

5. Cross-AZ deployment: even if the primary and standby nodes are in different data centers, the data can be replicated through Direct Connect in real time. If the local node is the primary and the remote node is the standby, the local node will be accessed first, and if it fails or becomes unreachable, the remote standby node will be upgraded to a primary node for service provision.

TDSQL-A for PostgreSQL supports the high availability solution based on strong sync replication. If the primary node fails, the system will automatically select the optimal standby node immediately to take over the tasks. The switch process is imperceptible to users, and the access IP remains unchanged. TDSQL-A for PostgreSQL offers 24/7 continuous monitoring for system components, and if there is a failure, it will automatically restart or isolate the failed node and select a new primary node from the standby nodes to continue service provision.

Features

TDSQL-A for PostgreSQL provides two data encryption methods:

Encryption on the business side

The business calls the encryption function built in TDSQL-A for PostgreSQL and writes encrypted results to the database. Normally, the encrypted data will be read and decrypted in the application.

Encryption built in TDSQL-A for PostgreSQL

The encryption process is transparent to the business, which has the following advantages:

1. Encryption operations (function calls) are decoupled from the business. The business is only responsible for writing the original data to the database kernel, and the subsequent encryption calculation will be conducted inside the database, which is imperceptible to the business.

2. Encryption algorithms are maintained by the database, and operations such as encryption algorithm selection and key management are all performed by the security admin independently.

Kernel-based encryption calculation supports async encryption to implement data encryption while delivering a stable system throughput. The supported encryption algorithms include AES-128, AES-192, AES-256, and SM4.

TDSQL-A for PostgreSQL supports column storage. You can define data tables as columnar tables based on your business needs. Generally, we recommend that you set tables with a large width and requirements for a high compression ratio as columnar tables. Columnar tables support a wide variety of compression algorithms, such as delta, zlib, zstd, RLE, and bitpack, which have different compression levels. For more information, please see the corresponding section in the user manual. TDSQL-A for PostgreSQL supports the new-generation columnar vectorized execution engine, which offers high query performance for hybrid row/column storage and query.

TDSQL-A for PostgreSQL supports transparent data masking that can return masked data to unauthorized users in an imperceptible manner.

TDSQL-A for PostgreSQL supports all-round auditing in multiple dimensions. Bypass detection is used for auditing, which has little impact on database operations. The following types of audits are available:

1. Statement audit: audits a certain type of statement.

2. Object audit: audits the operations of a certain database object.

3. User audit: audits the operations of a certain database user.

4. Fine-Grained audit (FGA): uses expressions as audit conditions and allows you to set actions for when an audit is triggered, such as sending emails or making phone calls.

The kernel natively supports hot/cold data separation so that the business can provide a unified database view without having to perceive the differences between the underlying storage media.

1. Hot data and cold data are stored in different node groups with different physical server configurations to implement hot/cold data separation and reduce costs.

2. Scheduled backend tasks can automatically migrate data according to the configured hot/cold data rules. This allows the system to implement automated hot/cold data separation and eliminate the need for the business to pay attention to the storage of hot/cold data in the cluster.

This feature is currently available on the Private Cloud Edition but unavailable on the Public Cloud Edition.

TDSQL-A for PostgreSQL ensures the cluster disaster recovery capabilities in multiple dimensions:

Strong Sync Replication

TDSQL-A for PostgreSQL supports strong sync replication. Ensuring that the primary and standby nodes have the identical data is the basis of the entire disaster recovery system. If the primary node fails, the database service can be switched to the standby node without any data loss. The strong sync replication mechanism requires that a success be returned only after the user request is executed and the log is successfully written to the standby node, guaranteeing that the data on the primary and standby nodes are always consistent.

Primary/Standby High Availability

The primary/standby high availability solution of TDSQL-A for PostgreSQL mainly uses multi-replica redundancy in each node group to ensure that there are no or only momentary service interruptions. If the primary node in a group fails and cannot be recovered, a new primary node will be automatically selected from the corresponding standby nodes to continue service provision. Based on primary/standby high availability, TDSQL-A for PostgreSQL supports the following features:

1. Automated failover: if the primary node in the cluster fails, the system will automatically select a new primary node from the corresponding standby nodes, and the failed node will be automatically isolated. The strong sync replication policy ensures complete primary/standby data consistency in case of primary/standby failover, fully meeting the finance-grade requirements for data consistency.

2. Failure recovery: if a standby node loses data due to disk failure, the database admin (DBA) can recover the standby node by building it again or add a standby server to a new physical node to recover the primary/standby relationship and thus ensure the system reliability.

3. Replica switch: each node in the primary/standby architecture (which can contain one primary node and multiple standby nodes) has a complete data replica that can be switched to by the DBA as needed.

4. Do-Not-Switch configuration: it can be set that failover will not be performed during the specified period of time.

5. Cross-AZ deployment: even if the primary and standby nodes are in different data centers, the data can be replicated through Direct Connect in real time. If the local node is the primary and the remote node is the standby, the local node will be accessed first, and if it fails or becomes unreachable, the remote standby node will be upgraded to a primary node for service provision.

TDSQL-A for PostgreSQL supports the high availability solution based on strong sync replication. If the primary node fails, the system will automatically select the optimal standby node immediately to take over the tasks. The switch process is imperceptible to users, and the access IP remains unchanged. TDSQL-A for PostgreSQL offers 24/7 continuous monitoring for system components, and if there is a failure, it will automatically restart or isolate the failed node and select a new primary node from the standby nodes to continue service provision.

Scenarios

TDSQL-A for PostgreSQL is suitable for real-time online analysis and offline analysis. It is commonly used as a pooling database to receive data from multi-source business systems and provide various data analysis-based query services, which support a wide variety of use cases such as multidimensional statistical analysis, smart modeling, commercial BI report analysis, and data resource management, as well as use cases with high data security requirements.

Pricing

No relevant information is provided as the product is currently unavailable for purchase.