Pricing policy for serverless mode Yandex Database

When using Yandex Database in serverless mode, a fee is charged for each request made to the database. You don't need to specify what resources are required, as the database quickly adapts to changes in the user load within the quotas allocated to the user. Apart from requests, the user pays for the data stored in Yandex Database on an hourly basis. Other operations, such as restoring data from backups, are charged additionally.

What goes into the cost of using serverless mode Yandex Database

When using Yandex Database in serverless mode, you pay for the following:

• Data operations.
• The amount of stored data, including service data, such as indexes.
• Additional user operations, like restoring data from backups.

Other consumed resources to be additionally paid for:

• The space used in Object Storage to store on-demand backups.
• Outgoing traffic from Yandex.Cloud to the internet.

In all calculations, 1 GB = 2¹⁰ MB = 2²⁰ KB = 2³⁰ bytes.

Data operations and the request unit

Yandex Database serverless operation mode supports multiple ways to work with data:

• Document API: Amazon DynamoDB-compatible HTTP API. You can use this API to perform operations on document tables.
• The YDB API and its implementations like the YDB CLI and YDB SDK for Java, Python, Node.js, and Go. You can use this API to perform operations on YDB tables.

To calculate the cost of requests in YDB, we use a concept called request units (RU). Each executed request, depending on its type, complexity, and data size, consumes a certain number of RU. The total cost of all executed requests to YDB is the sum of the cost of each request in RU.

Document API

The cost of requests to the Document API depends on the type of DB operations that are performed within the request and the amount of data that the operation is performed on. If the request consists of multiple operations (for example, BatchWriteItem), its cost is the sum of operation costs.

Operation type	Data size	Cost in request units (RU)
Item read operation (GetItem, BatchGetItem, Query, and Scan methods)	4 KB	1
Item read operation performed within a transaction (TransactGetItems)	4 KB	2
Item write operation (PutItem, BatchWriteItem, and UpdateItem)	1 KB	2
Item write operation performed within a transaction (TransactWriteItems)	1 KB	4
Item delete operation (DeleteItem)	ignored	2
DB schema operation (CreateTable, DeleteTable, DescribeTable, and ListTables)	Free	Free

When estimating the cost of a request:

• A read request that returned no result is always valued at 1 RU.
• An item write request that returned no results is always valued at 2 RU.
• An item write request within a transaction that returned no result is always valued at 4 RU.
• Read requests are rounded up to a multiple of 4 KB.
• Write requests are rounded up to a multiple of 1 KB.

YDB API

The cost of a request to the YDB API factors in the following:

• CPU usage time

  The CPU time used for parsing and compiling a request is taken into account. The CPU usage time is rounded down to a multiple of 1.5 ms.

• Read operations

  The cost of read operations is assessed three ways:

  • The total amount of data read. The amount is rounded up to a multiple of 4 KB.
  • Number of rows read.
  • The CPU time used for read operations. The CPU usage time is rounded down to a multiple of 1.5 ms.

  All values are converted to RU. The total cost of read operations is assumed to be equal to the largest of these values.

• Write operations

  The cost of write operations is assessed three ways:

  • The total amount of data written. The amount is rounded up to a multiple of 1 KB.
  • Number of rows written.
  • The CPU time used for write operations. The CPU usage time is rounded down to a multiple of 1.5 ms.

  All values are converted to RU. The total cost of write operations is assumed to be equal to the largest of these values.

• Delete operations

  The cost of delete operations is assessed three ways:

  • The total amount of data deleted. The amount is rounded up to a multiple of 1 KB.
  • Number of rows deleted.
  • The CPU time used for delete operations. The CPU usage time is rounded down to a multiple of 1.5 ms.

The total cost of the request is calculated as the sum of all components.

Estimated parameter	Unit size	Cost in request units (RU)
Amount of data read	4 KB	1
Amount of data written	1 KB	2
Amount of data deleted	1 KB	2
Item read operation	1 item	1
Item write operation	1 item	2
Item delete operation	1 item	2
CPU usage time	1.5 ms	4
Creating, updating, and deleting table schemas, getting a table description, getting a list of tables, creating and deleting directories	Free	Free

Specifics of cost calculation for certain operations

• ReadTable
  Операция ReadTable позволяет эффективно вычитывать большие диапазоны данных таблицы. Стоимость запроса зависит только от объема прочитанных данных и составляет 128 RU за 1 МБ. При расчете стоимости объем округляется в большую сторону до значения, кратного 1 МБ.

• BulkUpsert
  BulkUpsert позволяет эффективно загружать данные в базу. Стоимость записи строки в рамках операции BulkUpsert составляет 1 RU за 1 КБ записанных данных. При расчете стоимости объем данных округляется в большую сторону до значения, кратного 1 КБ. Итоговая стоимость операции рассчитывается как сумма стоимости всех записанных строк.

• Building a secondary index
  Стоимость построения индекса складывается из стоимости операции ReadTable оригинальной таблицы и BulkUpsert в индексную таблицу. Тарифицируется объем проделанной работы — количество строк и их размер, который был прочитан из оригинальной таблицы и записан в индексную таблицу. Процесс построения индекса может быть отменен пользователем. Отмененные операции построения индекса также тарифицируются за объем проделанной работы до отмены операции.

The amount of stored data

In serverless mode, data storage capacity is allocated automatically. The amount of stored data is calculated as the total amount of user and service data that is stored in the database. For example, creating a global index increases the storage size by the size of this index.

Creating backups

Automatic backups

Yandex Database automatically creates and stores two full backups for the last two days for free. No fee is charged for storing automatic backups.

On-demand backups

You can force a database backup with a copy saved in Yandex Object Storage. The cost of this operation depends on the amount of data copied. When calculating the cost, the actual amount is rounded up to a multiple of 1 GB.

Yandex Object Storage. The cost of storing backups is calculated according to the pricing policy for Yandex Object Storage.

Restoring data from backups

You can restore databases and individual tables from backups that are stored in Object Storage. The cost of this operation depends on the amount of data restored. When calculating the cost, the actual amount is rounded up to a multiple of 1 GB.

Discount for committed volumes of services (CVoS)

You can get a guaranteed discount on service resource usage when you plan for a month, a year, or three years ahead.

Yandex Database serverless mode provides a CVoS for a particular number of requests in RU per second. You can choose the committed volume in 100 RU/s increments.

When calculating the actual consumption, the number of requests is averaged every five minutes. That is, for example, if the committed volume is 100 RU/s, it isn't necessary to consume 100 RUs per second. Within the averaging interval, the load may change and amount to 150 RU/s during one minute and 50 RU/s another.

Pricing

All prices are shown with VAT. Prices are given for a 30-day month. For shorter months, the price is higher. For longer months, it's lower.

When using the service, you pay for traffic from Yandex.Cloud to the internet. Traffic between Yandex.Cloud services and incoming internet traffic are free.

The minimum billable unit is 1 MB.