AnalyticDB for PostgreSQL instance specifications and recommendations - AnalyticDB

This topic describes AnalyticDB for PostgreSQL instance specifications and provides recommendations.

Instance resource types

We recommend that you select the following AnalyticDB for PostgreSQL instance resource types:

Elastic storage mode
This instance resource type uses an integrated computing and storage architecture to provide comprehensive features. You can change compute node specifications, add compute nodes, and scale the storage capacity.
When you purchase an instance in elastic storage mode, you must specify the Edition, Compute Node Specifications, Nodes, Storage Disk Type, and Single Node Storage Capacity parameters.
Serverless mode
This instance resource type uses an in-house decoupled computing and storage architecture to implement on-demand storage and scaling within seconds. It is ideal for scenarios where resource requirements fluctuate significantly.
When you purchase an instance in Serverless mode, you must specify the Edition, Compute Node Specifications, and Nodes parameters.

The following tables describe these two instance resource types.

Table 1. Elastic storage mode

Edition	Node specifications	Recommended disk storage type	Suitable scenario
High-availability Edition	2 cores, 16 GB	PL0 Enterprise SSD (ESSD)	Proof of concept (POC) testing. Individual learning and a trial or testing of service features.
	4 cores, 32 GB	PL0 ESSD PL1 ESSD	Balanced computing and storage scenarios. This type of specifications is a choice for 60% of users.
	8 cores, 64 GB	PL1 ESSD	Compute-intensive scenarios, where large amounts of complex data are analyzed or concurrently queried.
	16 cores, 128 GB	PL2 ESSD	Construction of enterprise-class platforms, where large amounts of enterprise core data are concurrently queried.
High-performance Edition (Basic Edition)	2 cores, 8 GB	PL0 ESSD	POC testing. Individual learning and a trial or testing of service features.
	4 cores, 16 GB	PL0 ESSD PL1 ESSD	Balanced computing and storage scenarios for batch data analysis. Important High-performance Edition (Basic Edition) does not provide high availability. Proceed with caution.
	8 cores, 32 GB	PL1 ESSD
	16 cores, 64 GB	PL2 ESSD

Table 2. Serverless mode

Edition	Scheduling mode	Node specifications or AnalyticDB compute units (ACUs)	Disk storage type	Suitable scenario
High-availability Edition	Manual	4 cores, 16 GB	Shared storage	The new instance resource type can implement on-demand storage, data sharing, and node specification changes within seconds. It is suitable for the following scenarios: Resource requirements fluctuate significantly. New resource plans are not defined. Business workloads are distinctly isolated.
		8 cores, 32 GB
	Automatic	8 to 32 ACUs

Cases for selecting instance specifications

Case 1: Internet and manufacturing users

Internet and manufacturing users want to migrate data from self-managed databases and Greenplum data warehouses to the cloud.

Suggestion: We recommend that you select AnalyticDB for PostgreSQL in elastic storage mode.

Benefits: AnalyticDB for PostgreSQL is compatible with Greenplum, PostgreSQL, and other open source ecosystems. Data can be seamlessly migrated to AnalyticDB for PostgreSQL. After migration, resources can be adjusted based on business requirements.

Case 2: Internet SaaS users

Internet SaaS users need to build a data mid-end that runs stably. On this data mid-end, they want to perform extract-transform-load (ETL) operations on a variety of data sources such as ApsaraDB RDS, Realtime Compute for Apache Flink, and OSS, implement hybrid transactional and analytical processing (HTAP), and work with BI reports and enterprise-class data services.

Suggestion: We recommend that you select AnalyticDB for PostgreSQL in elastic storage mode. Recommended instance specifications: High-availability Edition, compute node specifications higher than 4 cores and 32 GB, and more than four compute nodes.

Benefits: The elastic storage mode allows data import from other Alibaba Cloud services or third-party cloud services. It provides enterprise-class capabilities, such as workload management based on user-defined functions or resource queues. The elastic storage mode supports ETL operations and provides computing performance about three times that of traditional data warehouses. It supports changes to applications that supply the data for analysis. It supports node specification changes and storage scaling based on business requirements.

Case 3: digital transformation of traditional enterprises

Traditional enterprises need to perform digital transformation and replace traditional data warehouses such as Teradata, Oracle, Db2, and Greenplum data warehouses on IDCs with cloud services.

Suggestion: We recommend that you select AnalyticDB for PostgreSQL in elastic storage mode. Recommended instance specifications: High-availability Edition or Basic Edition (based on your business requirements), compute node specifications higher than 4 cores and 32 GB, and more than four compute nodes.

Benefits: AnalyticDB for PostgreSQL is an industry-leading service that can replace Teradata and Oracle data warehouses. It has provided successful solutions for hundreds of financial institutions, Internet service providers (ISPs), public service sectors, and enterprises.

Case 4: autonomous driving enterprises

Autonomous driving enterprises need to perform geographical and time series analysis on the vehicle-collected data. They require JSON compatibility and spatio-temporal data analysis capabilities to build business dashboards and support feature engineering.

Suggestion: We recommend that you select AnalyticDB for PostgreSQL in elastic storage mode. Recommended instance specifications: Basic Edition and compute node specifications higher than 4 cores and 32 GB.

Benefits: The elastic storage mode supports the PostGIS and GanosBase engines for spatio-temporal analysis and can implement accelerated queries in the massively parallel processing (MPP) architecture. It supports flexible analysis on semi-structured data (such as JSON data) and data lake analysis.

Case 5: Internet gaming enterprises

Internet gaming enterprises need to build a data mid-end to analyze user behavior data. The data mid-end must provide business log cleansing and data join analysis to support gaming operations tools. Internet gaming enterprises need to process HTAP workloads and isolate resources during working hours.

Suggestion: We recommend that you select AnalyticDB for PostgreSQL in Serverless mode. Recommended instance specifications: compute node specifications higher than 4 cores and 16 GB and more than four compute nodes.

Benefits: The Serverless mode can adjust resources flexibly to meet business requirements during different time frames. It provides an efficient Log Service + OSS solution for log data cleansing. The Serverless mode has powerful analysis and single-node computing capabilities.

Case 6: new retail enterprises

New retail enterprises need to build a customer data platform (CDP) that is capable of importing data from multiple sources and selecting appropriate customers.

Benefits: The elastic storage mode supports a variety of data formats such as JSON, CSV, Avro, and Parquet to aggregate data and generate tags. It helps implement one-stop cloud-based platform building by working with other in-house services of Alibaba Cloud such as Quick Audience.

Case 7: large-scale Internet enterprises

Large-scale Internet enterprises have independent business mid-ends for each business unit and a unified data mid-end. They hope that independent resources can be efficiently deployed to support different business workloads and data silos are not generated in the future.

Benefits: The Serverless mode provides efficient resource deployment and elasticity to prevent heavy pre-planning and suit changing workloads. Data can be shared among multiple instances. This frees you from the concern of data silos that may be generated as a result of the development of business mid-ends and the data system construction. A single instance can completely isolate resources. This way, you can read separate resource usage of business units from the bill.

Case 8: building a data development platform

Users need to build a data development platform to reduce the impact of development on business and improve the development efficiency.

Benefits: AnalyticDB for PostgreSQL in Serverless mode supports the data sharing feature. When you need to perform data development, you can use this feature to consume the data shared by test instances in the production environment. This prevents the impact of development on the production environment and provides up-to-date data for development.