Community

Blog Events Webinars Tutorials Forum
  1. Blog
  2. Events
  3. Webinars
  4. Tutorials
  5. Forum
Create Account
×
Community Blog Seamless DB+AI Transformation: Why AnalyticDB for PostgreSQL Outshines Traditional Greenplum Solutions

Seamless DB+AI Transformation: Why AnalyticDB for PostgreSQL Outshines Traditional Greenplum Solutions

The article introduces the advantages of AnalyticDB for PostgreSQL compared to traditional Greenplum solutions, focusing on the seamless transformation of database and AI capabilities.

By Bo Pang

1. Overview

The cloud-native data warehouse AnalyticDB for PostgreSQL is fully compatible with Greenplum syntax, offering a fivefold performance improvement over open-source Greenplum. It provides a rich set of enterprise-level features, surpassing Greenplum in real-time computing, elastic scalability, enhanced security, and high availability. AnalyticDB for PostgreSQL extends data warehouse capabilities with vector retrieval and end-to-end RAG services, helping businesses to quickly build AI applications and enable integrated Data and AI capabilities.

2. Compatibility with Greenplum

2.1 Syntax Compatibility

• Support all complex data types and function computations available in Greenplum.

• Support 16 common expression calculations including CaseWhen, NullIf, and Coalesce.

• Support the scan of HeapTable, Aocs Table, and Beam Table, including Seqscan and Index Scan.

• Support GroupAgg, HashAgg, and PlainAgg, along with all Agg features, including AggFilter, GroupingSets, RollUp, and Cude.

• Support HashJoin and NestLoopJoin, and fully support seven Join rules of Left, Right, Full, Inner, Anti, Semi, and Not-exist-in.

• Support all Sort scenarios, including FullSort and TopNSort.

• Agg, Join, and Sort operations support storing data on disks.

• Window computing.

• Support Insert, Update, and Delete operations.

• Support custom functions and PROCEDUREs.

2.2 Ecosystem Compatibility

1. Compatibility with Community Tools

• Data Integration: Talend, Kettle, DSG, and others.

• Data Development: Navicat, DBeaver, and others.

• Job Scheduling: Informatica, Azkaban, and others.

• BI Analysis: Superset, Zeppelin, Grafana, FineBI, PowerBI, Tableau, Cognos, SmartBI, and others.

2. Compatibility with Alibaba Tools

• Data Integration: DTS, DataX, Flink, and others.

• Data Development and Scheduling: DMS, Dataworks, DataQ, Dataphin, and others.

• BI Analysis: QuickBI, DataV, and others.

3. Advantages Compared with Greenplum

3.1 Functional Advantages

AnalyticDB for PostgreSQL enhances product usability through proprietary enterprise-level features. Compared with Greenplum, it offers comprehensive advantages in data traces, real-time capabilities, security, reliability, and scalability, significantly reducing the use and O&M costs for businesses.

3.1.1 Zero-ETL

Support seamless integration of multiple data sources, including PolarDB and RDS, into AnalyticDB for PostgreSQL to build an enterprise-level data warehouse. Users can easily initiate full or incremental data integration from TP to AP directly in the AnalyticDB for PostgreSQL console without configuring data integration tools, significantly improving the integration performance of full data initialization.

3.1.2 Real-time Computing

Support high-throughput write operations for streaming data and provide incremental real-time materialized views, enabling multi-table joins between stream tables and offline tables, and real-time Ad-hoc queries. The same engine can handle both streaming and batch tasks by utilizing resource isolation to support mixed workloads, enabling businesses to build lightweight real-time data warehouses.

3.1.3 Enhanced Security

• Support TLS 1.1, 1.2, and 1.3.

• Support encryption of cloud disks.

• Support asymmetric and symmetric encryption, such as SM4.

• Support row-level and column-level access controls.

• Support dynamic data masking at the engine level.

• Support SQL auditing and event auditing.

3.1.4 Hot and Cold Data Tiering

AnalyticDB for PostgreSQL supports hot-cold tiered storage, enabling tables to be stored in OSS to reduce storage costs. It supports overall hot-cold conversion for data tables and hot-cold conversion by partition for partitioned tables, and also supports automatic data TTL (Time to Live).

3.1.5 Backup Recovery

Open-source Greenplum provides logical backups based on gpbackup, which requires configuring scheduled tasks and backup storage media. AnalyticDB for PostgreSQL offers distributed physical backup recovery based on consistent recovery points. This feature allows users to choose recovery at a specific point in time or instance cloning.

3.1.6 Cross-AZ Disaster Recovery

Cross-AZ deployment capabilities will be released in early July, enabling cross-AZ data disaster recovery to provide businesses with the most reliable data protection.

3.1.7 Elastic Scalability

Support vertical scaling of Master nodes and compute nodes, as well as horizontal scaling of compute nodes. During scaling, the system maintains read and write services and supports pausing and restarting scaling to avoid peak business hours.

3.1.8 Data Lake Acceleration

Support accelerated analysis of data in the data lake (OSS, MaxCompute) through the external table. Support high-speed integration of OSS data into AnalyticDB for PostgreSQL using ODPS FDW, and parallel high-speed import of MaxCompute data into AnalyticDB for PostgreSQL.

3.2 Performance Advantages

In terms of performance, AnalyticDB for PostgreSQL has developed features such as a vectorized execution engine and a hybrid row-column storage engine, real-time materialized views, result set caching, accelerated execution of the dictionary, and Dynamic Join Filter. With these proprietary performance-enhancing features, AnalyticDB for PostgreSQL's fully self-developed compute engine achieves a fivefold performance improvement over open-source Greenplum under identical resource conditions on the TPC-H 100GB benchmark.

1

For specific performance testing, please refer to the TPC-H performance testing for AnalyticDB for PostgreSQL V7.0.

3.3 Cost Advantages

  1. AnalyticDB for PostgreSQL provides customers with 8 CU of primary and secondary resources of Master nodes.
  2. After deeply integrated with the latest generation of AMD EPYC servers, AnalyticDB for PostgreSQL sees a 30% performance boost, achieving higher cost-effectiveness with the price unchanged. This integration is currently enabled by default in Beijing, Shanghai, Hangzhou, and Shenzhen. AnalyticDB for PostgreSQL continues to optimize performance to provide higher cost-effectiveness for customers.
  3. The deployment architecture of AnalyticDB for PostgreSQL features a single node with one primary and one secondary, ensuring the optimal cost-effective deployment mode while maintaining high availability.
  4. Support multi-level cloud disk storage, including PL0, PL1, and PL2, allowing customers to choose the most cost-effective option based on their performance requirements.
  5. Support hot-cold tiered storage, enabling low-frequency data to be stored through OSS with a significant reduction in costs while preserving most database usage patterns.
  6. Fully managed service, providing free O&M monitoring, along with kernel diagnostics and a proprietary high-performance engine. Compared with open-source solutions with equivalent performance, AnalyticDB for PostgreSQL can save up to 60% in resource consumption.
  7. Support scheduled start-stop and scaling operations for instances, enabling on-demand resource allocation. For users with significant business fluctuations in peak and off-peak periods, this can further reduce costs by over 30%.

4. Integrated Data and AI Capabilities

Based on practical experience across industries, combining large models with vector search engines and full-text engines to create an RAG architecture has become the most controllable, efficient, and data-driven technical solution for deploying AI applications.

AnalyticDB for PostgreSQL introduces a new approach of integrating Data and AI, extending its data warehouse capabilities to support tag filtering, vector search, and full-text search for integrated analysis. It also provides an end-to-end RAG service (document processing, embedding, recalling, and refinement) within the warehouse, avoiding data silos and complex maintenance caused by introducing multiple engines for AI.

Different from other products, AnalyticDB for PostgreSQL is integrated with Alibaba Cloud Model Studio, DingTalk, and Alibaba Cloud PAI. It also provides enterprise-specific knowledge bases, Chatbot, text/image search, and other solutions for enterprises to directly use in building AI applications.

5. One-Click Migration from Greenplum

If you want to gain the above advantages to solve pain points related to performance, functionality, and management, and to achieve an architectural upgrade and transformation towards integrated Data and AI, migrate to AnalyticDB for PostgreSQL. The following migration solution helps you migrate data to AnalyticDB for PostgreSQL efficiently and cost-effectively and enable integrated Data and AI capabilities. For more information, see Migrate data from a self-managed Greenplum cluster to an AnalyticDB for PostgreSQL instance.

Database AI Big Data Data Warehouse Open Source cloud native Greenplum AnalyticDB for PostgreSQL AI Applications Retrieval-Augmented Generation RAG
0 1 0
Share on

Read previous post:

Alibaba Cloud DTS Experience Sharing Series | Guide to Accelerating Full Data Migration

Read next post:

[Infographic] Highlights | Database New Features in September 2024

ApsaraDB

451 posts | 96 followers

You may also like

Comments

ApsaraDB

451 posts | 96 followers

Related Products

More Posts by ApsaraDB

See All