Skip to main content

EKS Container Cost

A page for analyzing EKS Pod costs. It supports two data sources: OpenCost (default) or Request-based estimation (fallback).

EKS Container Cost

Key Features​

Data Source Indicator​

The current data source is displayed at the top of the page:

  • Green: OpenCost (Prometheus) - Actual usage based, CPU + Memory + Network + Storage + GPU
  • Yellow: Request-based estimation - CPU + Memory only (OpenCost installation recommended)

Stats Cards​

  • Pod Cost (Daily): Total daily Pod cost (cyan)
  • Pod Cost (Monthly): Estimated monthly cost (green)
  • Running Pods: Running Pod count / Node count (purple)
  • Top Namespace: Highest cost namespace (orange)

Namespace Cost Distribution Chart​

Pie chart showing daily cost distribution by namespace

Node Daily Cost + Pod Count Chart​

Dual-axis bar chart showing daily cost and Pod count per node

Pods Tab​

ColumnDescription
NamespaceNamespace
PodPod name
NodeNode name
CPUCPU cost
MemoryMemory cost
Network*Network cost (OpenCost only)
Storage*Storage cost (OpenCost only)
GPU*GPU cost (OpenCost only)
Total/DayTotal daily cost

*Displayed only in OpenCost mode

Nodes Tab​

ColumnDescription
NodeNode name
Instance TypeEC2 instance type
Hourly RateHourly cost
Daily CostDaily cost
PodsPod count

Two Cost Calculation Methods​

Method A: Request-based (Default)​

Distributes node cost by Pod resource request ratio:

CPU Ratio = Pod CPU Request / Node Allocatable CPU
Memory Ratio = Pod Memory Request / Node Allocatable Memory
Pod Daily Cost = (CPU Ratio x 0.5 + Memory Ratio x 0.5) x Node Hourly Rate x 24h

Supported items: CPU, Memory only Data source: Steampipe kubernetes_pod, kubernetes_node

Method B: OpenCost (Prometheus)​

Combines actual usage metrics with AWS pricing information:

CPU Cost = Actual CPU Usage (cores) x AWS EC2 vCPU Price
Memory Cost = Actual Memory Usage (bytes) x AWS EC2 Memory Price
Network Cost = Cross-AZ/Region Transfer x Data Transfer Price
Storage Cost = PVC Provisioned Size x EBS Volume Price
Pod Total Cost = CPU + Memory + Network + Storage + GPU

Supported items: CPU, Memory, Network, Storage, GPU (5 items) Data source: Prometheus + Metrics Server

OpenCost Installation​

bash scripts/07-setup-opencost.sh

After installation, setting opencostEndpoint in data/config.json automatically switches to OpenCost mode.

How to Use​

  1. Click Compute > EKS Container Cost in the sidebar
  2. Check the data source from the banner at the top
  3. Review overall cost status from the stats cards
  4. Identify high-cost namespaces/nodes from the charts
  5. Switch between Pods/Nodes tabs to check detailed costs
  6. Expand the "Cost Calculation Basis" section to verify calculation basis

EC2 Pricing Reference (ap-northeast-2, On-Demand)​

Instance TypeHourly Rate
m5.large$0.118
m5.xlarge$0.236
m6g.large$0.100
c5.xlarge$0.196
r5.large$0.152
t3.large$0.104
t4g.large$0.086

Tips​

OpenCost Installation Recommended

Request-based only considers resource requests and may differ from actual usage. Installing OpenCost enables accurate analysis of 5 cost items.

Pods Without Requests

Pods without resource requests show as $0.00 in Request mode. As a best practice, set resource requests for all Pods.

Network Cost (OpenCost)

OpenCost's Network cost includes only Cross-AZ transfers. Same-AZ transfers are free.

AI Analysis

You can analyze with the AI Assistant using queries like "EKS Pod cost analysis", "Cost comparison by namespace", "Cost optimization recommendations", etc.