This topic describes the billable items and billing rules of AnalyticDB for PostgreSQL.
Usage notes
The prices that are provided in this topic are for reference only. See the service console to obtain the actual prices of individual resources.
Coordinator node CUs
Billing method | Fee calculation | Example |
Pay-as-you-go | You are billed based on the specifications and usage duration of coordinator node compute units (CUs). Coordinator node CU fee = Unit price per coordinator node CU × Usage duration Note
| You purchase a pay-as-you-go AnalyticDB for PostgreSQL instance in the China (Beijing) region. The instance uses 16 CUs of coordinator node resources. The instance is used for 1,500 hours. The coordinator node CU fee is calculated by using the following formula: USD 4.166 per hour × 1,500 hours = USD 6,249. |
Subscription | You purchase a subscription AnalyticDB for PostgreSQL instance in the China (Beijing) region. The instance uses 16 CUs of coordinator node resources. The subscription period is two months. The coordinator node CU fee is calculated by using the following formula: USD 2133.18 per month × 2 months = USD 4,266.36. |
Computing resources
Elastic storage mode
Edition | Billing method | Fee calculation | Example |
High-performance Edition (Basic Edition) |
| You are billed based on the compute node specifications, the number of compute nodes, and the usage duration. Computing resource fee = Unit price per compute node × Number of compute nodes × Usage duration |
|
High-availability Edition |
Serverless mode
Scheduling mode | Billing method | Fee calculation | Example |
Manual | Pay-as-you-go | You are billed based on the compute node specifications, the number of compute nodes, and the usage duration. Computing resource fee = Unit price per compute node × Number of compute nodes × Usage duration | You purchase a pay-as-you-go AnalyticDB for PostgreSQL instance in the China (Beijing) region. The instance uses four compute nodes that each provide 2 cores and 8 GB memory. The instance is used for 2,000 hours. The computing resource fee is calculated by using the following formula: USD 0.186 per hour × 4 × 2,000 hours = USD 1,488. |
Automatic | Pay-as-you-go | You are billed based on the number of AnalyticDB compute units (ACUs) and the usage duration. Computing resource fee = Unit price per ACU × Number of ACUs × Usage duration | You are not charged for computing resources during public preview of automatic scheduling. |
Storage resources
Elastic storage mode
Edition | Storage type | Billing method | Fee calculation | Example |
High-performance Edition (Basic Edition) | PL1 enhanced SSD (ESSD) | Same as the billing methods for computing resources | You are billed based on the storage type, the storage size per node, the number of compute nodes, and the usage duration. Storage resource fee = Unit price of storage × Storage size per node × Number of compute nodes × Usage duration |
|
High-availability Edition |
|
AnalyticDB for PostgreSQL supports tiered storage of hot and cold data. The cold data that is stored in Object Storage Service (OSS) incurs storage costs. For more information, see Tiered storage of hot and cold data.
Serverless mode
Billing method | Fee calculation |
Pay-as-you-go | You are billed based on the storage actually occupied (including basic data and backup data) and the usage duration. Storage resource fee = (Basic data size + Backup data size) × Unit price of storage × Usage duration |
RAG service
The RAG service of AnalyticDB for PostgreSQL uses API operations to address the limitations of general large language models (LLMs) in AI-generated content (AIGC) scenarios. You are charged based on the number of tokens when you use the embedding and rerank models. The number of tokens is returned in the response of the API operations.
Billing method | Fee calculation | Example |
Pay-as-you-go | You are billed based on the number of tokens. RAG service fee = Number of tokens × Unit price per token | You purchase an AnalyticDB for PostgreSQL instance in elastic storage mode in the China (Beijing) region. The vector search engine optimization feature is enabled for the instance. When you use the bge-reranker-v2-m3 model to call the Rerank operation, the number of tokens returned in the response is 2,000. The RAG service fee is calculated by using the following formula: USD 0.00131577727293801 per 1,000 tokens × 2,000/1,000 = USD 0.0026315. |