Understanding split cost allocation data
Split cost allocation data splits the cost of a shared AWS resource across the workloads and resources that use it. The shared resource's own line item keeps its total usage and cost, and additional line items show how the usage and cost are allocated in your Cost and Usage Reports (AWS CUR).
For column definitions, see Split line item details for legacy CUR and Split line item columns for CUR 2.0.
Split cost allocation data for Amazon ECS and Amazon EKS
Using split cost allocation data, you can allocate your container costs to individual business units and teams, based on how your container workloads consume shared compute and memory resources. Split cost allocation data introduces cost and usage data for new container-level resources (that is, ECS tasks and Kubernetes pods) to AWS CUR. Previously, AWS CUR only supported costs at the EC2 instance level. Split cost allocation data generates container-level costs by looking at each containerâs EC2 instance resource consumption, and generates cost based on the amortized cost of the instance and the percentage of CPU and memory resources consumed by the containers that ran on the instance.
For accelerated computing instances used with Amazon EKS, split cost allocation data
includes resource allocation for specialized processors alongside CPU and memory. This
covers NVIDIA and AMD GPUs, AWS Trainium, and AWS Inferentia accelerators. The
feature is available only for Amazon EKS environments and provides pod-level resource
reservation data for these accelerated computing resources. This allows you to track and
allocate costs for workloads that use these specialized processors, such as AI/ML
applications and other computationally intensive tasks. For a current list of
accelerated computing instances, see Accelerated
Computing
To receive split cost allocation data for containers, turn on Amazon ECS, Amazon EKS, or both in Cost Management preferences, and configure your legacy CUR or CUR 2.0 export to include split cost allocation data. For setup instructions, see Enabling split cost allocation data.
Split cost allocation data introduces new usage records and new cost metric columns for each containerized resource ID (that is, ECS task and Kubernetes pod) in AWS CUR. For more information, see Split line item details.
Split cost allocation data for networking
For networking, split cost allocation data allocates supported NAT gateway, Amazon EC2 instance, and VPC peering connection charges to the resources that generated the usage. A networking flexible cost allocation policy specifies which charges to allocate and the reporting granularity, such as by Amazon EC2 instance, VPC, or AWS account. For policy setup and supported Regions, see Allocate shared networking costs with networking flexible cost allocation in the Amazon Virtual Private Cloud User Guide.
To receive networking split cost allocation data, associate a networking flexible cost allocation policy with the resource and turn on showback. Select Networking flexible cost allocation in Cost Management preferences, and configure your legacy CUR or CUR 2.0 export to include split cost allocation data and resource IDs. For setup instructions, see Enabling split cost allocation data.
Split cost allocation data introduces new usage records for each resource that generated networking usage and for unattributed usage in AWS CUR. For more information, see Example of split cost allocation data for networking and Split line item details.
Split cost allocation data for Amazon S3
For Amazon S3, split cost allocation data divides the storage and request charges of a general purpose bucket among the prefixes in that bucket. It allocates storage charges for all storage classes and Tier 1 and Tier 2 request charges.
Amazon S3 Storage Lens publishes prefix-level storage and request metrics to Billing and Cost Management, and you choose which buckets and prefix levels to publish in an S3 Storage Lens dashboard configuration. To receive this data, enable Billing and Cost Management publishing on an S3 Storage Lens advanced tier dashboard, select Amazon S3 in Cost Management preferences, and configure your legacy CUR or CUR 2.0 export to include split cost allocation data and resource IDs. For setup instructions, see Enabling split cost allocation data.
Split cost allocation data introduces a usage record for each prefix that has usage and for unattributed usage in AWS CUR. For more information, see Example of split cost allocation data for Amazon S3, and Split line item details for legacy CUR and Split line item columns for CUR 2.0.
Cost allocation tags
For ECS: When it comes to AWS cost allocation tags, you can use Amazon ECS-managed tags or user-added tags for your Cost and Usage Reports. These tags apply to all new ECS split cost allocation data usage records. For more information, see Tagging your ECS resources for billing.
For EKS: Split cost allocation data creates new cost allocation tags for some
Kubernetes attributes. These tags include aws:eks:cluster-name,
aws:eks:deployment, aws:eks:namespace,
aws:eks:node, aws:eks:workload-name, and
aws:eks:workload-type.
aws:eks:cluster-name,aws:eks:namespace, andaws:eks:nodeare populated retrospectively with the name of the cluster, namespace, and node.aws:eks:workload-typeis only populated if there is exactly one workload managing the pod, and is one of the built in workloads. Workload types includeReplicaSet,StatefulSet,Job,DaemonSet, orReplicationController, andaws:eks:workload-nameincludes the name of the workload. For more information, see Workloadsin the Kubernetes Documentation. aws:eks:deploymentis only populated for the workload typeReplicaSet. It is the deployment that creates aReplicaSet.
These tags apply to all new EKS split cost allocation data usage records. These
tags are enabled for cost allocation by default. If you previously used and disabled
the aws:eks:cluster-name tag, then split cost allocation data keeps
this setting and doesn't enable the tag. You can enable it from the Cost allocation tags
For networking: Split cost allocation data adds the following AWS-generated tags:
aws:networking-fca:allocated-accountidentifies the AWS account that owns the resource that generated the usage. On the unattributed line item, its value is(unattributed).aws:networking-fca:allocation-policyidentifies the allocation policy and version, for examplefcap-0123456789abcdef0:2. On the unattributed line item, it's empty.
These tags apply to allocated and unattributed networking split cost allocation data usage records.
For Amazon S3: Split cost allocation data adds the following AWS-generated tags to prefix line items:
aws:s3:PrefixLevel1throughaws:s3:PrefixLevel15hold the segments of the prefix path through the deepest matched level. On the unattributed line item,aws:s3:PrefixLevel1is(unattributed).aws:s3:BucketNameholds the bucket name.
These tags are inactive by default. To include them in your reports, activate them on the Cost allocation tags
Estimating the number of line items
When including split cost allocation data in AWS CUR, two new usage records are added for each ECS task and Kubernetes pod per hour in order to reflect the CPU and memory costs. To estimate the number of new line items in AWS CUR per day, use the following formula:
For ECS: (number of tasks * average task lifetime * 2) * 24
For EKS: (number of pods * average pod lifetime * 2) * 24
For example, if you have 1,000 pods running each hour across a cluster of 10 EC2 instances and the lifetime for the pod is less than 1 hour, then:
(1000 * 1 * 2) * 24 = 48,000 new usage records in AWS CUR
For accelerated computing instances in Amazon EKS, three new usage records are added for each Kubernetes pod per hour in order to reflect the accelerator, CPU, and memory costs. To estimate the number of new line items in AWS CUR per day, use the following formula:
For EKS with accelerated computing: (number of pods * average pod lifetime *
3) * 24
For example, if you have 1,000 pods running each hour across a cluster of 10 EC2
instances and the lifetime for each pod is less than one hour, then: (1000 * 1
* 3) * 24 = 72,000 new usage records in AWS CUR
For each usage type and operation that the policy allocates on a shared networking resource, split cost allocation data adds one line item for each resource that generated usage.
To estimate the number of new line items in AWS CUR per day, use the following formula:
For networking: (number of shared networking resources * number of usage types and operations the policy allocates per resource * number of resources that generated usage) * 24
For example, if 200 instances use one NAT gateway and the policy allocates one usage type and operation for the gateway, the report adds (1 * 1 * 200) * 24 = 4,800 line items per day. Reporting at a broader level, such as VPC or account, reduces the number of line items.
For Amazon S3, split cost allocation data adds daily prefix line items for storage and requests. To estimate the number of new line items that Amazon S3 adds to AWS CUR in a month, use the following formula:
For Amazon S3: (number of prefixes with usage + 1) * number of usage types with usage (up to 2) * days in the month
For example, a bucket with 2,000 prefixes at the deepest level in your range that has storage and request usage every day adds up to (2,000 + 1) * 2 * 30 = 120,060 line items in a 30-day month. Prefix line items are daily, not hourly. A prefix with no usage of a type on a day has no line item for that type.