Billing
This document outlines the billing for OpenSearch HA3 engine instances.
Billing methods
Method | Description |
Subscription | You pay when you create an instance. This method costs less than pay-as-you-go. Subscription instances cannot be released early. You cannot change a subscription instance to a pay-as-you-go instance. |
Pay-as-you-go | A bill is generated for your usage each hour. All hourly bills are part of a single order, and the fees are deducted from your Alibaba Cloud account. You can change the billing method from pay-as-you-go to subscription. If the 'Convert to Subscription' button is not available in the console, submit a ticket. |
Pricing (General fees)
OpenSearch bills you for each application instance. Six instance types are available for different scenarios. Fees are calculated based on instance rental, storage capacity, and compute resources. Compute resources are measured in Logic Compute Units (LCUs). The total fee is calculated as follows: Total fee = Instance rental fee + Storage capacity fee + Compute resource fee. To calculate your specific costs, you can use the pricing calculator.
Subscription
Specification | Scenarios | Quota range | Purchase step size | Instance rental (CNY/month) | Price per LCU (CNY/month) | Price per GB of storage (CNY/month) |
Shared General-purpose | For scenarios that require fast search responses and have large data volumes. | Compute resources: [20 LCU, 5,000 LCU]. Storage capacity: [1 GB, 300 GB]. | 1 LCU, 1 GB | 229.21 | 1.37 | 52.38 |
Shared Compute-optimized | For scenarios that require very fast search responses and have small data volumes. | Compute resources: [1,000 LCU, 20,000 LCU]. Storage capacity: [1 GB, 50 GB]. | 1 LCU, 1 GB | 229.21 | 1.53 | 58.45 |
Shared storage types | For scenarios that have large data volumes and do not require fast search responses. | Compute resources: [5 LCU, 200 LCU]. Storage capacity: [5 GB, 500 GB]. | 1 LCU, 1 GB | 229.21 | 1.77 | 21.82 |
Dedicated General-purpose | Dedicated resources with better performance than the Shared General-purpose type. | Compute resources: [500 LCU, 200,000 LCU]. Storage capacity: [100 GB, 2,000 GB]. | 500 LCU, 100 GB | 271.38 | 4.6 | 145.66 |
Dedicated Compute-optimized | Dedicated resources with better performance than the Shared Compute-optimized type. | Compute resources: [1,000 LCU, 500,000 LCU]. Storage capacity: [10 GB, 500 GB]. | 1,000 LCU, 10 GB | 271.38 | 5.19 | 335.11 |
Dedicated Storage-optimized | Dedicated resources with better performance than the Shared Storage-optimized type. | Compute resources: [500 LCU, 2,000 LCU]. Storage capacity: [500 GB, 2,000 GB]. | 500 LCU, 500 GB | 254.77 | 1.38 | 28.25 |
Pay-as-you-go
Specification | Scenarios | Quota range | Purchase step size | Instance rental (CNY/hour) | Price per LCU (CNY/hour) | Price per GB of storage (CNY/hour) |
Shared General-purpose | For scenarios that require fast search responses and have large data volumes. | Compute resources: [20 LCU, 5,000 LCU]. Storage capacity: [1 GB, 300 GB]. | 20 LCU, 1 GB | 0.42 | 0.0026 | 0.096 |
Shared Compute-optimized | For scenarios that require very fast search responses and have small data volumes. | Compute resources: [1,000 LCU, 20,000 LCU]. Storage capacity: [1 GB, 50 GB]. | 1,000 LCU, 1 GB | 0.42 | 0.0028 | 0.107 |
Shared Storage | For scenarios that have large data volumes and do not require fast search responses. | Compute resources: [5 LCU, 200 LCU]. Storage capacity: [5 GB, 500 GB]. | 5 LCU, 5 GB | 0.42 | 0.0033 | 0.041 |
Dedicated General-purpose | Dedicated resources with better performance than the Shared General-purpose type. | Compute resources: [500 LCU, 200,000 LCU]. Storage capacity: [100 GB, 2,000 GB]. | 500 LCU, 100 GB | 0.49 | 0.0083 | 0.263 |
Dedicated Compute-optimized | Dedicated resources with better performance than the Shared Compute-optimized type. | Compute resources: [1,000 LCU, 500,000 LCU]. Storage capacity: [10 GB, 500 GB]. | 1,000 LCU, 10 GB | 0.49 | 0.0094 | 0.605 |
Dedicated Storage-optimized | Dedicated resources with better performance than the Shared Storage-optimized type. | Compute resources: [500 LCU, 2,000 LCU]. Storage capacity: [500 GB, 2,000 GB]. | 500 LCU, 500 GB | 0.46 | 0.0025 | 0.051 |
Note:
For more information about pricing, see the Pricing overview.
When you view application details or create an application using an API or SDK, the quota specifications use the following English names: Shared Compute-optimized (opensearch.share.compute), Shared Storage-optimized (opensearch.share.storage), Shared General-purpose (opensearch.share.common), Dedicated Compute-optimized (opensearch.private.compute), Dedicated Storage-optimized (opensearch.private.storage), and Dedicated General-purpose (opensearch.private.common).
Bill item descriptions: For pay-as-you-go instances, the billable items are: consumption.portal.opensearch.spec/buytype: Instance rental (CNY/hour), consumption.portal.opensearch.doc_size/storage: Price per GB of storage (CNY/hour), and consumption.portal.opensearch.compute_resource/compute: Price per LCU (CNY/hour). For subscription instances, the billable items are: buytype: Instance rental (CNY/month), storage: Price per GB of storage (CNY/month), and compute: Price per LCU (CNY/month).
Industry templates are free of charge in all regions during the public preview period. You will be charged for the templates after they are officially launched.
Pricing (Model training fees)
OpenSearch provides model services such as custom analyzers. Models that are not listed in the following table are in public preview and are free of charge.
Models do not incur instance rental fees. The storage capacity they consume is included in the total storage of the instance.
If a model is referenced in a search request, the compute resources consumed by the search are included in the total compute resources of the instance.
Each time a model is trained, you are charged on a pay-as-you-go basis for the actual billable hours consumed. This applies to both subscription and pay-as-you-go instances.
Model type | Storage capacity | Compute resources | Model training price (CNY/billable hour) | Estimated billable hours for model training |
Text analyzer model | Each version of an analyzer model uses 1 GB of storage resources. | Each search request consumes a fixed 2 LCU, plus 0.01 LCU for each search result returned. | 3.87 | Training with the minimum amount of data consumes 250 billable hours. To increase training resources, submit a ticket to contact technical support. |
Term weighting model | Each version of a term weighting model uses 1 GB of storage resources. | Each search request consumes 10 LCU. | 3.87 | Training with the minimum amount of data consumes 250 billable hours. To increase training resources, submit a ticket to contact technical support. |
Note:
Billable hours = Model training duration (hours) × Number of CPU cores used for training
Selecting an instance type
For more information about the differences between dedicated and shared instance types, see Select an OpenSearch edition.
Select a quota: The storage quota is based on the actual size of your data in bytes in OpenSearch. To estimate your initial storage quota, you can multiply the number of documents by the average document size. For your initial LCU quota, you can start with the minimum number of LCUs for your chosen instance type. After you create the application, you can monitor its LCU usage in Cloud Monitor during testing and scale out as needed. Upgrades and downgrades take effect immediately without interrupting your search service.
Introduction to LCU compute resources: An LCU is a unit of measurement for search computing power. One LCU represents the computing power of 0.01 CPU cores in a search cluster.
Method for estimating compute resources: Number of LCUs = QPS × LCUs consumed per query. You can determine the number of LCUs consumed per query using the search test feature.