Improve cost visibility of Amazon ECS and AWS Batch with AWS Split Cost Allocation Data

Shubir Kapoor, Principal Product Manager, contributed to this blog

We’re excited to announce that the cost data for Amazon Elastic Container Service (Amazon ECS) tasks and AWS Batch jobs is now available in the AWS Cost and Usage Reports (CUR). With AWS Split Cost Allocation Data, you can easily understand and optimize cost and usage of your containerized applications, and allocate application costs back to individual business entities based on how shared compute and memory resources are consumed by your containerized applications. Split Cost Allocation Data for ECS tasks and AWS Batch jobs is available to customers in all AWS commercial regions, excluding China regions, with no additional cost.

The challenges of monitoring and allocating container costs

As you modernize your applications with Amazon ECS, a fully managed container orchestration service, you increase efficiency by consolidating and maximizing resource utilization to deploy and scale your applications. You can use AWS Batch to schedule and execute your containerized jobs in Amazon Elastic Compute Cloud (Amazon EC2) and AWS Fargate environments that dynamically provision and scale compute resources based on the volume and requirements of your submitted jobs. However, the efficiency that comes with sharing compute resources makes it difficult to distribute your application costs at the container task level. Additionally, the dynamic nature of containerized platforms makes it even more complex to understand and allocate your container costs (e.g., individual containers can use more resources than the reserved CPU; memory capacity or workloads can be rescheduled across instances/availability zones/regions).

In order to optimize and chargeback the cost of your containerized applications, you need more granular cost data, e.g., ECS task cost. Prior to the availability of Split Cost Allocation Data, you could access the cost of individual billable resources, such as EC2 instances, but didn’t have visibility into granular resources (such as ECS tasks and AWS Batch jobs), since these container task-level resources were not used to compute your bill. You could potentially calculate the cost of the ECS resources with EC2 cost data and the usage data of ECS deployed on EC2 from your CUR report. Nevertheless, the process is complex and requires significant investment.

Get granular cost visibility of ECS tasks and AWS Batch jobs with Split Cost Allocation Data

With Split Cost Allocation Data, you can now easily distribute your Amazon EC2 instance costs at the ECS task level, based on the actual consumption of the vCPU and memory utilized by your ECS tasks and your AWS Batch jobs. The granular cost information at the container task level lets you analyze the cost efficiency of your containerized applications and simplifies the chargeback process to your business entities.

After Split Cost Allocation Data is enabled, it will scan for all of your ECS tasks or AWS Batch jobs across your Consolidated Billing family and ingest usage telemetry data (actual and reserved CPU and memory) for your ECS resources and compute new ECS task-level metrics, e.g., split cost, unused cost, actual usage, and reserved usage, including net cost metrics after discounts. If you tag ECS resources or AWS Batch jobs for billing purposes (see ECS developer guide and AWS Batch user guide for details) and enable these tags for cost allocation, Split Cost Allocation Data will ingest ECS-managed tags and user-added tags to the CUR. You can use these tags, as well as AWS Cost Categories, to better share container-level cost and usage data at the desired business entity level, e.g., team and application level. Meanwhile, the availability of unused cost data helps identify unused compute or memory resources, enabling opportunities to optimize your container cluster and minimize inefficiencies.

Get started with Split Cost Allocation Data

To enable “Split cost allocation data”, you need to complete a simple two-step opt-in process. First, as a payer account owner, you need to opt-in “Split cost allocation data” from your AWS Cost Management/AWS Cost Explorer preference page. Then, you will enable “Split cost allocation data” for a new or existing CUR report from the Cost and Usage Reporting preference page in the AWS Billing console. Once enabled, the report will automatically scan for ECS tasks for the entire Consolidated Billing family (all clusters belonging to member accounts across all regions) and start preparing the granular cost data for the current month. In 24 hours, your CUR report will be ready with the new ECS container task level metrics.

Figure 1. Enable “Split cost allocation data” from Cost Explorer preferences

Figure 2. Configure Cost and Usage Report (CUR) for “Split cost allocation data”

How it works

To help you understand how the cost model works, we’ll start by explaining a few key terms and calculation logic. Split Cost Allocation Data for ECS collects the reserved and actual usage data of compute and memory resources for each EC2 instance associated with an ECS cluster. It then calculates the allocated CPU and memory data for each ECS task based on the greater value between the reserved amount and the used amount.

Figure 3. Allocated resources as the greater of reserved and actual utilization

Since multiple ECS tasks can run on a single EC2 instance, Split Cost Allocation Data for ECS computes the  allocated CPU and memory for all tasks on the instance. It then computes a split-usage-ratio, the percentage of vCPU or memory allocated to each ECS task compared to the overall CPU or memory available on the EC2 instance. It also identifies the unused capacity left on the instance.

Figure 4. Resources allocated to multiple ECS tasks

When sharing the cost of the EC2 instance among all the ECS tasks, Split Cost Allocation Data totals the split cost and then proportionally redistributes the unused cost according to instance utilization by task.

Figure 5. Total allocated cost equals split cost plus proportionally redistributed unused cost

Example: EC2 cost allocation across multiple ECS tasks

Now, let’s plug in some numbers and show how an EC2 instance cost is allocated across multiple ECS tasks.

Below you can see an ECS cluster run on a single node of m7g.2xlarge EC2 instance with 8 vCPU and 32GB RAM. The cluster is operating 4 ECS tasks across 2 ECS services. We will use the on-demand pricing for m7g.2xlarge for demonstration purposes.  Split Cost Allocation Data uses relative unit weights for CPU and memory based on a 9:1 ratio. This is derived from average relative price of vCPU to memory based on Fargate pricing. Using these weights, it computes the cost per vCPU-Hour and cost per GB-Hour as $0.029 and $0.003, respectively.

Table 1. m7g.2xlarge EC2 instance

The vCPU and memory reservation and actual usage across the 4 ECS tasks are listed in the table below. Task 2 used more CPU and memory than what was reserved because it didn’t configure a limit. As mentioned earlier, Split Cost Allocation Data computes allocated vCPU and memory based on the greater value between the reserved and actual usage. In this example, we don’t have any unused vCPU, but there is 2GB of unallocated memory.

Table 2. Reserved, used, and allocated data for the ECS cluster

Split Cost Allocation Data computes a split-usage ratio as the percentage of CPU or memory allocated by the ECS task compared to the overall CPU or memory available on the EC2 instance. It also computes an unused ratio as the percentage of CPU or memory allocated by the ECS task compared to the overall CPU or memory allocated on the EC2 instance (that is, not factoring in the unallocated CPU or memory on the instance). For instance, there is 2GB of unallocated memory on the instance, which is 0.063 of the total instance memory. Therefore, the total unused memory ratio for Task 1 is 0.188 / (1-0.063) = 0.200.

The task-level split costs are calculated based on split-usage ratios multiplied by the cost per vCPU-Hour and cost per GB-Hour. If there is unused resource, in this case unused memory capacity of 2GB, the unused instance cost ($0.006) is proportionally distributed to each task based on the unused ratio computed for each task. The total allocated cost for each task will be the total of the split cost and proportionally redistributed unused cost. Once the EC2 cost is available at the task level, you can calculate the aggregate service-level costs. In this example, the total allocated cost for ECS service 1 and ECS service 2 are $0.142 and $0.188, respectively. You can also aggregate the costs using tags or Cost Categories that were added to the ECS tasks, allowing you to compute the cost by your desired business entity level.

Table 3 Allocated cost data at the ECS task level

What are the new Cost and Usage Report columns?

As Split Cost Allocation Data computes ECS task-level metrics, you will see the new columns in your CUR reports. For instance, “SplitUsage” stands for the usage for vCPU or memory allocated for the specified time period to the Amazon ECS task. You can view this user guide for a complete list of the new CUR columns and their definitions.

To wrap up with our example, below is a demo CUR report.  You can see how data will be displayed in the new CUR columns.

Table 4. Sample Cost & Usage Report data

Conclusion

The visibility of cost data at the ECS container-task level provides insights to improve efficiency and cost optimization. Enable this granular level of cost data today and learn more about this capability.