Reserved read throughput and reserved write throughput are data table settings available for high-performance instances. Both are measured in capacity units (CU). Tablestore uses these settings to guarantee the specified throughput for each data table.
Configuration
When you create a data table in a high-performance instance, you can specify reserved read throughput, reserved write throughput, or both.
The configured values represent the read and write capacity that Tablestore guarantees per second. For example, if you set reserved read throughput to 100 CU, Tablestore guarantees 100 read CU per second.
Reserved read throughput and reserved write throughput each accept any value from 0 CU up to a default cap of 100,000 CU. The two caps are applied independently.
If you need throughput beyond the default cap, submit a ticket.
To adjust the reserved read/write throughput after the data table is created, see Update a data table.
Billing
Any non-zero reserved read or write throughput incurs charges, regardless of actual request volume.
Every read or write request made to a data table through the Tablestore API consumes read or write capacity units, respectively.
Billing unit
Reserved read throughput and reserved write throughput are measured in read capacity units (read CU) and write capacity units (write CU) respectively.
Billing cycle
Tablestore bills the total reserved read and write throughput across all data tables in a high-performance instance hourly.
Reserved throughput settings can change within an hour. Tablestore samples each data table's reserved read/write throughput at fixed intervals, averages the samples over the hour, and multiplies the hourly average by the unit price.
Unit price
For the unit price of reserved read/write throughput, see Tablestore pricing.