Frequently asked questions about Astra DB PCUs
This page answers common questions about Provisioned Capacity Units (PCU) in Astra DB. For general questions about Astra DB Serverless, see Frequently asked questions about Astra DB Serverless.
What is a PCU?
A Provisioned Capacity Unit (PCU) is a specific amount of computing capacity that your Astra DB Serverless databases can use to support complex, latency-sensitive workloads or accommodate periodic usage spikes.
With PCUs, you decide how much capacity you need and which databases can use that capacity.
How do PCUs compare to the default capacity and scaling potential?
Both PCU and standard capacity have a minimum capacity and potential for autoscaling additional capacity.
Databases that use standard capacity attempt to mirror actual database usage as much as possible. Standard capacity automatically scales up and down based on demand.
With PCUs, you decide how much capacity you want your databases to have by default, regardless of demand. You can also set the maximum capacity for autoscaling to support unexpected increases in demand.
Additionally, PCUs are designed to handle more complex workloads with greater throughput compared to standard capacity. They also offer additional customizations like dedicated provisioning and cache optimization.
How much does a PCU group cost?
PCU pricing depends on the PCU subscription options, which are defined in the PCU group settings, and actual hourly usage. For more information, see PCU subscriptions.
What are RCUs and HCUs? What is the difference between reserved capacity and minimum capacity?
Reserved Capacity Units (RCUs) and Hourly Capacity Units (HCUs) are billing categories for PCUs.
When you create a PCU group, you define the reserved, minimum, and maximum capacity. These values determine the amount of PCUs allocated to the group. Depending on the group’s configuration, the PCUs can be billed at the RCU rate or the HCU rate.
For much more information, see Capacity allocation.
PCU group configuration
The following questions relate to creating and editing PCU groups.
For more information and instructions, see Create PCU groups in Astra DB and Edit PCU groups in Astra DB.
Can I put multiple databases in a PCU group?
Yes, but the group’s allocated capacity must be able to support those databases. For more information about distributing databases across PCU groups, see Plan your PCU groups in Astra DB.
Why can’t I add a database to my PCU group?
There are several reason a database can’t be added to a PCU group:
-
The database already belongs to another PCU group.
-
The database’s deployment type doesn’t match the other databases in the PCU group. A PCU group can have either Serverless (Non-Vector) or Serverless (Vector) databases, but not both.
-
The database isn’t deployed to the same cloud provider and region as the PCU group.
If none of the above apply to your database, contact DataStax Support to investigate the issue.
What are some examples of invalid PCU group settings?
-
The minimum capacity can’t be less than the reserved capacity.
-
The maximum capacity can’t be less than the minimum capacity.
-
Reserved, minimum, and maximum can’t be zero, except for flexible capacity workloads where the reserved capacity must be zero.
-
The group can’t have a mix of Serverless (Non-Vector) and Serverless (Vector) databases. Each group can have only one type of database.
Can I change the number of RCUs or HCUs in a PCU group?
Yes, but there are limitations. For more information, see Edit PCU groups in Astra DB and Capacity allocation.
How do I disable autoscaling in a PCU group?
To disallow autoscaling for a PCU group, set the maximum capacity equal to the minimum capacity in the PCU group’s settings.
PCU usage
The following questions relate to PCU usage, including scaling and fluctuating resource consumption.
Is it possible to have HCUs that run constantly?
Yes. See Hourly Capacity Units (HCUs) and Committed and flexible workloads.
How long does it take additional capacity (HCUs) to scale up?
If database demand exceeds a PCU group’s minimum capacity, autoscaling begins provisioning HCUs up to the group’s maximum capacity to accommodate the additional demand. It can take up to 15 minutes for the additional capacity to be fully provisioned and available.
Likewise, when you unpark a PCU group, it can take time to scale up the group’s minimum HCUs and reactivate the hibernated databases.
What happens if demand surpasses the maximum capacity?
If demand exceeds a PCU group’s maximum capacity, you can experience issues with database availability and performance.
If this is a frequent occurrence, consider increasing the group’s allocated capacity.
For more information about capacity allocation, see Plan your PCU groups in Astra DB.
How can I ensure that a PCU group can meet additional demand? How do I know when to add additional capacity?
Monitor your PCU usage to determine if your groups are using their allocated capacity efficiently or need additional capacity. Based on the usage metrics and your predicted future usage, you can draw conclusions about the group’s capacity needs.
For more information about capacity allocation and handling fluctuating traffic, see Plan your PCU groups in Astra DB.
Is there a limit to the number of RCUs or HCUs per group? What if I need more units?
The maximum possible capacity is 10 units.
If your PCU group has multiple databases, DataStax recommends moving some databases into other PCU groups to distribute the load and avoid exceeding the maximum possible capacity.
If you need more than 10 units for a single database, contact your DataStax account representative or DataStax Support.
For capacity allocation guidance, see Plan your PCU groups in Astra DB.
