All Products
Search
Document Center

Tablestore:Billable items of search indexes

Last Updated:Jul 28, 2023

Search index-based queries require extra space to store indexed data and increase read throughput. This topic describes billable items, billing formulas, and billing examples of search indexes.

Usage notes

  • Billable items of indexes are independent of data tables.

  • The price of each billable item of a search index is the same as the price of each billable item in a high-performance instance. For more information about pricing, see Tablestore Pricing.

Billable items

Billable item

Billing method

Description

Data size

Pay-as-you-go

Unit: GB. If the data size is not an integer, the data size is rounded up to an integer.

Tablestore charges you for the total volume of indexed data on an hourly basis. The utilization of system resources varies based on the field types and index types. You are charged for indexed data based on the volume of data that is compressed after you create indexes. The volume of raw data in a data table does not affect the billing.

Read throughput

Reserved read throughput

Pay-as-you-go

Unit: CU.

Tablestore automatically specifies a reserved read throughput based on the indexed data size. The fees for reserved read throughput are calculated based on the following operations:

  • When you create a search index, Tablestore reads data from the data table for which the search index is created. This process increases read throughput.

  • Tokenization also increases read throughput when you create a search index. Fees that are generated for tokenization are included in the bill that is generated for the reserved read throughput.

  • To ensure index performance and query performance, some indexed data is loaded to the memory in advance and remains in the memory. This increases read throughput. Fees that are generated based on the read throughput are also included in the bill that is generated for the reserved read throughput.

If the read throughput during search index-based queries is smaller than or equal to the reserved read throughput, you are charged based on the unit price of reserved read throughput. For example, the reserved read throughput is 10,000 CUs for an index, each index-based query reads 10 rows, and the size of each row is less than 4 KB. If the number of queries per second (QPS) is less than 1,000, the actual read throughput that is calculated is less than the reserved read throughput and no extra fees are generated.

Calculation of reserved read throughput: The reserved read throughput is proportional to the size of the indexed data and the number of rows of the indexed data. For example, 1 GB or 2 million rows of indexed data correspond to a reserved read throughput of 10 CUs. If the reserved read throughput value that corresponds to the data size is different from the reserved read throughput value that corresponds to the number of rows, the system uses the larger value as the reserved read throughput.

Important

If the data size is less than 200 MB and the number of rows is less than 400,000, the reserved read throughput for Tablestore must be 20 CUs. This value is suitable for tests that are performed on a small volume of data. If the data size is greater than or equal to 200 MB, or the number of rows is greater than or equal to 400,000, the reserved read throughput must be at least 100 CUs.

Metered read throughput

Pay-as-you-go

You are charged for the portion of the actual read throughput that exceeds the reserved read throughput based on the pay-as-you-go billing method.

Unit: CU.

Outbound traffic over the Internet

Pay-as-you-go

You are charged for the outbound traffic that is sent over the Internet. Unit: GB.

Billing formulas

The following table describes how to calculate the data size and read throughput of a search index.

Billable item

Formula

Description

Data size

fig_tablestore_001

The Size variable specifies the size of the compressed indexed data.

Read throughput

Reserved read CUs per index:fig_reservereadcu

Read CUs per query: cuperquery

  • The Size variable specifies the size of the compressed indexed data.

  • The Rows variable specifies the total number of rows in an index, excluding the child rows in nested fields.

  • The ReturnRowSize variable specifies the size of rows that are returned.

  • The ReturnRowCount variable specifies the number of rows that are returned.

Billing examples

Note

The prices in the following table are provided only for reference. For more information, see Tablestore Pricing.

Storage volume

Number of rows

Billing

8 GB

9 million

  • Storage fees: 8 GB × USD 0.00030 per GB-hour = USD 0.0024 per hour

  • Calculation of reserved read throughput: 8 GB of data corresponds to 80 CUs and 9 million rows correspond to 45 CUs. If the data size is greater than 200 MB and the number of rows is greater than 400,000, the system sets the reserved read throughput to a value that is greater than or equal to 100 CUs. Therefore, the value of 100 CUs is used as reserved read throughput in calculation.

    Fees of reserved read throughput: 100 CUs × USD 0.0002 per CU-hour = USD 0.02 per hour

  • Total fees: USD 0.0024 per hour + USD 0.02 per hour = USD 0.0224 per hour

You are charged for the portion of actual read throughput that exceeds the reserved read throughput based on the pay-as-you-go billing method. The fees for outbound traffic that is sent over the Internet in search indexes are the same as the fees for outbound traffic that is sent over the Internet in data tables.

100 GB

300 million

  • Storage fees: 100 GB × USD 0.00030 per GB-hour = USD 0.03 per hour

  • Calculation of reserved read throughput: 100 GB of data corresponds to 1,000 CUs and 300 million rows correspond to 1,500 CUs. The reserved read throughput is 1,500 CUs.

    Fees of reserved read throughput: 1,500 CUs × USD 0.0002 per CU-hour = USD 0.3 per hour

  • Total fees: USD 0.03 per hour + USD 0.3 per hour = USD 0.33 per hour

You are charged for the portion of actual read throughput that exceeds the reserved read throughput based on the pay-as-you-go billing method. The fees for outbound traffic that is sent over the Internet in search indexes are the same as the fees for outbound traffic that is sent over the Internet in data tables.

FAQ

Why are reserved read CUs generated when I use a search index?

Tablestore automatically specifies a reserved read throughput based on the indexed data size. The fees for reserved read throughput are calculated based on the following operations:

  • When you create a search index, Tablestore reads data from the data table for which the search index is created. This process increases read throughput.

  • Tokenization also increases read throughput when you create a search index. Fees that are generated for tokenization are included in the bill that is generated for the reserved read throughput.

  • To ensure index performance and query performance, some indexed data is loaded to the memory in advance and remains in the memory. This increases read throughput. Fees that are generated based on the read throughput are also included in the bill that is generated for the reserved read throughput.

Can I modify the reserved read CU settings for a search index?

No. You cannot modify the reserved read CU settings for a search index. The number of reserved read CUs is positively related to the data size and the number of rows. Reserved read CUs are used when you create a search index and for memory maintenance. If you want to reduce the number of reserved read CUs, you can reduce the size of the index or the number of rows.

Why am I charged for high-performance storage and high-performance read and write CUs when I use a search index in a capacity instance?

When you use a search index, extra space is required to store indexed data and search index-based queries increase read throughput. The price of each billable item of a search index is the same as the price of each billable item in a high-performance instance. Therefore, when you use a search index in a capacity instance, you are charged for high-performance storage and high-performance read and write CUs.

How do I view metering data of search indexes?

You can view metering data such as the storage volume, reserved read CUs, and number of rows of search indexes in the Tablestore console. For more information, see View the metering data of search indexes.

Can I use reserved CU plans to offset the fees that are generated for reserved read throughput of search indexes?

Yes. You can use reserved CU plans to offset the fees that are generated for reserved read and write CUs, including the reserved read throughput of search indexes.