- Newest
- Most votes
- Most comments
When you enable Split Cost Allocation Data (SCAD) for Amazon ECS in your Cost and Usage Reports (CUR), you can indeed allocate EC2 instance costs to individual containerized workloads based on their resource consumption.
The split cost allocation data generates container-level costs by analyzing each container's EC2 instance resource consumption based on the amortized cost of the instance and the percentage of CPU and memory resources consumed by the containers running on the instance.
For correlating container costs with their parent EC2 instances, you should understand that SCAD introduces new usage records and cost metric columns for each containerized resource ID (ECS tasks) in your CUR. For each ECS task, two new usage records are added per hour to reflect the CPU and memory costs.
Regarding which columns to use for accurate correlation:
The split cost allocation data is calculated by computing the cost of individual ECS services and tasks running on the cluster. The system looks at each container's EC2 instance resource consumption and generates cost based on the amortized cost of the instance and the percentage of CPU and memory resources consumed.
While the documentation doesn't explicitly state which specific columns (splitLineItem/PublicOnDemandSplitCost and splitLineItem/PublicOnDemandUnusedCost) should be used for container costs, the split cost allocation data is designed to ensure that the sum of all container costs (split costs) plus unused costs should equal the total cost of the parent EC2 instance.
To enable this feature, note that only regular and payer accounts have access to Cost Management preferences and can opt into Split Cost Allocation Data. After enabling, it can take up to 24 hours for the data to be visible in your CUR.
Sources
Cost Allocation Data for Amazon ECS | AWS re:Post
Understanding split cost allocation data - AWS Data Exports
COST03-BP01 Configure detailed information sources - Cost Optimization Pillar
It is still not clarified which split line item costs and unused costs correspond to which original EC2 cost. More specifically, which split line item columns sums to the unblended costs of a parent EC2 instance?
Relevant content
asked 2 years ago
asked 2 years ago
asked 8 months ago
- AWS OFFICIALUpdated a year ago
- AWS OFFICIALUpdated 2 months ago

Use
splitLineItem/PublicOnDemandSplitCostandsplitLineItem/PublicOnDemandUnusedCostfor ECS task costs. The sum should closely match the EC2 instance’s unblended cost, but expect minor differences. There is no 100% perfect correlation due to AWS’s allocation logic.There is indeed no good way to do what you are trying to do . This is because of a few reasons
Good luck trying to use SCAD. You can read more about the challenges here : https://medium.com/@sree-chalasani/why-aws-split-cost-allocation-data-is-unusable-aeef2f85e943