OCPU management and billing

Learn about Oracle Compute Units (OCPUs) and how you're billed for usage.

Note:

Customers using features in released in Limited Availability, you're not billed for usage of these resources, however, you must be aware of how you'll be billed when Limited Availability ends.

Metering and billing

Metering and billing for OCI GoldenGate is based on the number of OCPUs the service uses per minute.

When you create an OCI GoldenGate deployment, you select the number of OCPUs between 1 and 24 that your deployment will use. Each OCPU allocates an additional 16 GB memory.

You can enable auto scaling, which allows the service to scale up to three times the OCPU Count you specified upon creation. When you enable auto scaling, you are billed for the actual average number of OCPUs consumed per hour.

For example, if you specify 3 as the base OCPU and enable Auto Scale, then the total OCPUs that can be used is 9. When the OCPU Utilization is greater than 33.333% of 9 OCPUs, you are billed for the integer value over 33.333%, which is 4 OCPUs.

OCPU Utilization Greater than OCPU Utilization Less than or equal to Billed for
0 33.333% 3 OCPUs
33.333% 44.444% 4 OCPUs
44.444% 55.555% 5 OCPUs
55.555% 66.666% 6 OCPUs
66.666% 77.777% 7 OCPUs
77.777% 88.888% 8 OCPUs
88.888% 100% 9 OCPUs

OCI GoldenGate cannot scale over 24 OCPUs, which is the maximum number of OCPUs. For example, if you select 9 OCPUs as your base OCPU Count and enable Auto Scale, the service will scale up to 24 OCPUs. You can enable or disable auto scaling at any time.

You can monitor the OCPU consumption and memory of a deployment in the Metrics section of a deployment's Detail page.

About OCPU utilization

Before you create an OCI GoldenGate deployment, it's important to understand how the OCPU shape affects the OCI GoldenGate user experience.

When you create an OCI GoldenGate deployment, you must select how many OCPUs the deployment will use. Here are some things to consider when making your selection:

  • Enabling Auto Scale ensures additional OCPUs are available when they're needed. You're only billed when the additional OCPU cycles are consumed.
  • The OCI GoldenGate Backup process also consume OCPU cycles. If the current OCPU utilization is near 100%, the backup process can take a long time to complete.
  • If an OCI GoldenGate deployment is running at or near 100% OCPU utilization, consider changing the base OCPU count or enabling Auto Scale, if it's not already enabled.
  • You can monitor OCPU Utilization on the OCI GoldenGate Deployment Details page and the Metrics Explorer in the OCI Console.
  • You can set alarms in the OCI Console to notify you when OCPU utilization reaches certain levels.

Sizing considerations for Data Replication deployments

Ensure that you review the information in Metering and billing for OCI GoldenGate deployments about Oracle Compute Unit (OCPU) selection and scaling.

OCI GoldenGate Data Replication OCPU usage is calculated based on the following factors:

  • GoldenGate deployment console: If the current OCPU utilization is near 100%, then the GoldenGate Deployment Console may be very slow to return or not return at all. Oracle recommends allocating 1 OCPU for the user interface.
  • Number of Extracts, Replicats, and Distribution Paths processes running concurrently: Extract and Replicat processes are high priority processes and take precedence in OCPU cycles. Oracle recommends that you allocate 1 OCPU for each process.
  • Amount of data being processed

While there are many other factors that impact performance, Oracle recommends starting with the following configurations:

  • Development or Testing: 1 OCPU with auto-scaling enabled
  • Production: 4 OCPUs with auto-scaling enabled

After some testing, either keep the existing configuration or scale the deployment up or down manually.

Sizing considerations for Data Transforms deployments

Ensure that you review the information in Metering and billing for OCI GoldenGate deployments about Oracle Compute Unit (OCPU) selection and scaling.

OCI GoldenGate Data Transforms OCPU usage is calculated based on the following factors:

  • Data Transforms console
  • Number of Data Loads, Data Flows, and Workflows

While there are many other factors that impact performance, Oracle recommends starting with the following configuration:

  • Development or Testing: 1 OCPU with auto-scaling enabled
  • Production: 2 OCPUs with auto-scaling enabled

After some real world testing, you can either keep the existing configuration or scale the deployment up or down manually.

Sizing considerations for Stream Analytics deployments

Ensure that you review the information in Metering and billing for OCI GoldenGate deployments about Oracle Compute Unit (OCPU) selection and scaling.

OCI GoldenGate Stream Analytics OCPU usage is calculated based on the following factors:

  • Stream Analytics console
  • Number of Streaming pipelines
  • Ignite cluster
  • GoldenGate Big Data cluster

Before calculating the number of OCPUs you need, let's first review how many compute units each Stream Analytics resource requires. 1 OCPU is equal to 2 compute units (vCPUs). 1 vCPU is equal to 1000 millicores (1000m).

The following table lists example Stream Analytics pipeline settings and the calculated number of OCPUs required.
Pipeline Driver Executor Total vCPUs OCPUs billed
Pipeline A 500m 1 x 500m 1000m 1
Pipeline B 500m 2 x 500m 1500m 1
Pipeline C 500m 4 x 500m 2500m 2
Pipeline D 600m 2 x 700m 2000m 1
Pipeline E 1000m 2 x 1000m 3000m 2

You can configure the Driver and Executor settings as needed for each pipeline in the Stream Analytics console.

The following table lists example Stream Analytics resource configurations based on the number of pipelines (from the above table) and the calculated number of OCPUs required.

Stream Analytics console Number of pipelines Streaming pipelines Ignite cluster GoldenGate for Big Data cluster OCPUs billed
1000m 1 x Pipeline A 1000m 0 0 1
1000m 3 x Pipeline A 3000m 0 0 2
1000m 1 x Pipeline B 1500m 0 0 2
1000m 1 x Pipeline B 1500m 2 x 500m 500m 2
1000m 1 x Pipeline A

1 x Pipeline B

 2500m 2 x 500m 500m 3
1000m 2 x Pipeline A

1 x Pipeline B

 3500m 2 x 500m 500m 3

The Stream Analytics console requires 1000m. Each streaming pipeline requires additional millicores depending on their settings. The Ignite cluster, if activated, requires a minimum of 2 cluster instances. You can configure the millicore limit for both Ignite and GoldenGate Big Data clusters in the Stream Analytics console. When added together, you can determine the total number of OCPUs that you need to select when creating your Stream Analytics deployment.

If you're unsure, you can start with 2 or 3 OCPUs, and then review the OCPU consumption metrics on the deployment details page and adjust accordingly.