Mapping between instance "node" specifications and original "compute group" specifications

更新时间:
复制 MD 格式

On August 23, 2019, AnalyticDB for PostgreSQL changed how instances are defined. Instances moved from compute groups to compute nodes. This page maps old compute group types to their equivalent compute node types so you can confirm that your instance configuration, performance, capacity, and price are unchanged.

Background

The change from compute group to compute node simplified Massively Parallel Processing (MPP) cluster type selection and aligned the product with standard cluster database naming conventions.

The key structural difference:

  • Each compute node contains exactly one MPP data partition.

  • A compute group could contain multiple MPP data partitions depending on its type.

Because the total number of MPP data partitions stays the same across the mapping, the overall compute and storage capacity is identical — which means performance, capacity, and price are all equivalent.

Compute node to compute group mappings

The following table maps each legacy compute group type to its equivalent compute node configuration.

Column descriptions:

  • Compute group type (no longer used): The legacy instance type, retired on August 23, 2019.

  • Compute node type: The current instance type that replaces the compute group.

  • Number of MPP data partitions: The total number of parallel processing partitions across all nodes in the configuration. This count is identical for both the old and new type, which is why performance is equivalent.

  • Total amount of resources: The combined CPU cores, memory, and storage across all nodes. User storage is the space available for your data; total storage includes system overhead.

Compute group type (no longer used)Compute node typeNumber of MPP data partitionsTotal amount of resources
1x2C SSDTwo 1-core SSD nodes22-core/16 GB memory/160 GB user storage/320 GB total storage
1x16C SSDSixteen 1-core SSD nodes1616-core/128 GB memory/1,280 GB user storage/2,560 GB total storage
4x4C SSDFour 4-core SSD nodes416-core/128 GB memory/1,280 GB user storage/2,560 GB total storage
2x2C HDDTwo 2-core HDD nodes24-core/32 GB memory/2 TB user storage/4 TB total storage
4x4C HDDFour 4-core HDD nodes416-core/96 GB memory/8 TB user storage/16 TB total storage

Compute nodes and compute groups have the same performance, capacity, and price when they share the same amount of resources and same number of partitions.