Pricing
AI Data Platform Unit (AIDP Unit) is the unit of measurement used to meter consumption in your AI Data Platform.
AIDP Unit is a measure of the AI, data management, and data processing work done in your AI Data Platform instance. The rate at which AIDP Units are charged depends on the use of OCPU and memory resources per hour.
You can see how different cluster configurations are priced by using our Pricing Widget.
Table 2-1 AI Data Platform Units Mapping
AI Data Platform Definition | AIDP Units Charged |
---|---|
AMD OCPU per Hour (with included 100 GB Block storage per OCPU) | 67 AIDP Units |
ARM OCPU per Hour (with included 100 GB Block storage per OCPU) | 27 AIDP Units |
Intel OCPU per Hour (with included 100 GB Block storage per OCPU) | 87 AIDP Units |
NVIDIA GPU per Hour (with included Block Volume, CPU, Memory) | 4110 AIDP Units |
AMD Memory GB per hour | 3 AIDP Units |
ARM Memory GB per hour | |
Intel Memory GB per hour |
You pay AIDP Units to AI Data Platform for all your compute clusters and features based on the definitions in the table above. For object storage, logging, metrics, or any databases on OCI, you pay directly to OCI.
You select Compute Shape options between AMD, Intel, ARM, and NVIDIA GPUs. More specific or granular shapes are not currently available. Your selection of compute shapes is subject to the availability of compute shapes in the OCI region where your AI Data Platform instance is located. Oracle reserves the right to change the specific compute shapes available in an OCI region.
Compute clusters are a combination of compute OCPU and memory. When you create a compute cluster in AI Data Platform, you see an estimated cost of that compute cluster per hour in AIDP Units. For example:
- The cost Per Hour for an AMD 2 OCPU 32 GB Memory cluster will be 2 OCPU x 67 AIDP Units + 32 GB Memory x 3 AIDP Units = 230 AIDP Units.
- The cost Per Hour for an Intel 4 OCPU 32 GB Memory cluster will be 4 OCPU x 87 AIDP Units + 32 GB Memory x 3 AIDP Units = 444 AIDP Units.
- The cost Per Hour for an ARM 4 OCPU 32 GB Memory cluster will be 4 OCPU x 27 AIDP Units + 32 GB Memory x 3 AIDP Units = 204 AIDP Units.
- The cost Per Hour for a NVIDIA GPU cluster with 2 GPU will be 2 GPU x 4110 AIDP Units = 8220 AIDP Units.
By default, one Master Catalog Default Cluster is present in each AI Data Platform instance. This cluster is responsible for the essential AI Data Platform functions, like search crawls, refreshing catalog objects, creating, editing, and deleting objects, testing connections etc. Default Master Catalog Compute Clusters have AMD 2 OCPU 32 GB Memory. For the Default Master Catalog Compute Cluster, you are charged hourly for a minimum of 230 AIDP Units. Deleting an AI Data Platform instance halts the ongoing cost of that Default Master Catalog Compute Cluster. If you add more capacity to the Default Master Catalog Compute cluster to meet your performance and scalability needs, the AIDP U cost of the Default Master Catalog scales up relative to the increase.
Sample Pricing Scenario
In this scenario, you have 2 compute clusters running for a month (all calculations are in USD):
- 1 Default Master Catalog Compute Cluster: AMD 2 OCPU with 32 GB Memory
- 1 custom Cluster for workloads: Intel 4 OCPU with 32 GB Memory
In this case, the cost calculation for your AMD Cluster (for Default Master Catalog Compute Cluster) is:
- 2 OCPUs 32 GB Memory Per Hour = (2*67 + 32*3) AIDP Units =230 AIDP Units
- 230 AIDP Units/hour = 171,210 AIDP Units/month
- $0.230/hour = $171.21/month
The cost calculation for your Intel Cluster (for workloads) is:
- 4 OCPUs 32 GB Memory Per Hour = (4*87 + 32*3) AIDP Units = 444
- AIDP Units 444 AIDP Units/hour = 330,370 AIDP Units/month
- $0.444/hour = $330.37/month
Adding your monthly costs together, you total monthly costs would amount to: $171.21/month + $330.37/month = $501.58/month.