Kubernetes cluster cost estimate
Say how much CPU and memory your pods ask for, and see what the same cluster costs each month on AKS, EKS, GKE and OKE, using each vendor's public list prices. A managed Kubernetes cluster: the cluster fee, worker nodes sized from the CPU and memory the pods ask for (or pods billed directly, with no nodes), a load balancer, persistent volumes, and traffic out to users on the internet.
Step 1 of 3: Workload
Estimated monthly costIndia (Mumbai / Pune)
₹39,623a month
Oracle Cloud Lowest
Lowest of 4 at these inputs, 53.3% below the next.
Monthly total by provider
Each bar is one provider's monthly total, split into the rows of the table below. The same colour marks the same row everywhere.
| Provider | HTTP load balancer | Internet data transfer | Managed Kubernetes cluster fee | Block storage (general-purpose SSD) | General-purpose VM | Monthly total |
|---|---|---|---|---|---|---|
| Oracle Cloud | 6,931.35 (17%) | 807.08 (2%) | 31,884.21 (80%) | 39,622.64 | ||
| Amazon Web Services | 2,112.54 (2%) | 4,151.21 (5%) | 6,931.35 (8%) | 1,731.89 (2%) | 70,006.64 (82%) | 84,933.63 |
| Google Cloud | 2,302.78 (2%) | 5,697.00 (6%) | 6,931.35 (7%) | 2,278.80 (2%) | 80,850.04 (82%) | 98,059.96 |
| Microsoft Azure | 19,875.14 (19%) | 4,557.60 (4%) | 6,931.35 (7%) | 2,005.34 (2%) | 70,006.64 (68%) | 1,03,376.07 |
Estimated monthly cost
| Service line | Oracle Cloud | Amazon Web Services | Google Cloud | Microsoft Azure |
|---|---|---|---|---|
| Managed Kubernetes cluster fee | 6,931.35Oracle Kubernetes Engine (Enhanced cluster) | 6,931.35Amazon EKS | 6,931.35Google Kubernetes Engine (Standard, regional cluster) | 6,931.35Azure Kubernetes Service (Standard tier) |
FormulaOracle CloudOracle Kubernetes Engine (Enhanced cluster)
Amazon Web ServicesAmazon EKS
Google CloudGoogle Kubernetes Engine (Standard, regional cluster)
Microsoft AzureAzure Kubernetes Service (Standard tier)
| ||||
| General-purpose VM | 31,884.21OCI Compute (VM.Standard.E5.Flex)5 × E5.Flex 2 OCPU, 16 GB | 70,006.64Amazon EC2 (M6i)5 × m6i.xlarge (4 vCPU, 16 GB) | 80,850.04Google Compute Engine (N2)5 × n2-standard-4 | 70,006.64Azure Virtual Machines (Dsv5)5 × D4s v5 (4 vCPU, 16 GB) |
FormulaOracle CloudOCI Compute (VM.Standard.E5.Flex)
5 × E5.Flex 2 OCPU, 16 GB Amazon Web ServicesAmazon EC2 (M6i)
5 × m6i.xlarge (4 vCPU, 16 GB) Google CloudGoogle Compute Engine (N2)
5 × n2-standard-4 Microsoft AzureAzure Virtual Machines (Dsv5)
5 × D4s v5 (4 vCPU, 16 GB) | ||||
| Pod-billed Kubernetes compute | Not needed | Not needed | Not needed | Not needed |
FormulaOracle CloudNot needed Amazon Web ServicesNot needed Google CloudNot needed Microsoft AzureNot needed | ||||
| HTTP load balancer | 0.00OCI Flexible Load Balancer10 Mbps | 2,112.54Application Load Balancer0.82 LCUs on average | 2,302.78Cloud Load Balancing (global external Application Load Balancer) | 19,875.14Azure Application Gateway (Standard v2)0.84 capacity units on average |
FormulaOracle CloudOCI Flexible Load Balancer
10 Mbps Amazon Web ServicesApplication Load Balancer
0.82 LCUs on average Google CloudCloud Load Balancing (global external Application Load Balancer)
Microsoft AzureAzure Application Gateway (Standard v2)
0.84 capacity units on average | ||||
| Block storage (general-purpose SSD) | 807.08OCI Block Volume (Balanced) | 1,731.89Amazon EBS (gp3) | 2,278.80Compute Engine balanced persistent disk | 2,005.34Azure managed disks (Standard SSD, LRS) |
FormulaOracle CloudOCI Block Volume (Balanced)
Amazon Web ServicesAmazon EBS (gp3)
Google CloudCompute Engine balanced persistent disk
Microsoft AzureAzure managed disks (Standard SSD, LRS)
| ||||
| Internet data transfer | 0.00OCI outbound data transfer | 4,151.21AWS data transfer out | 5,697.00Google Cloud network egress (Premium Tier) | 4,557.60Azure bandwidth (Microsoft network routing) |
FormulaOracle CloudOCI outbound data transfer
Amazon Web ServicesAWS data transfer out
Google CloudGoogle Cloud network egress (Premium Tier)
Microsoft AzureAzure bandwidth (Microsoft network routing)
| ||||
| Monthly total | 39,622.64Lowest | 84,933.63+114.4% | 98,059.96+147.5% | 1,03,376.07+160.9% |
Assumptions
- Region
- India (Mumbai / Pune)
- Currency
- INR
- Prices
- On-demand list prices
- Hours a month
- 730
- Not included
- Support plans, taxes, negotiated discounts
What this workload means
- Worker nodes
- 5 (CPU needs 5, memory needs 5, minimum 3)
- Requests per month
- 52,560,000
- Data out to users
- 500 GB
- Load balancer traffic
- 0.82 GB an hour
How the quantities were worked out
- Prices are on demand, for 730 hours a month, with the cluster running all month.
- Nodes are sized from the CPU and memory the pods request. A node keeps part of its CPU and memory for the system, using the allocatable amounts Google publishes for GKE on every cloud; the other clouds reserve similar amounts.
- The cluster has at least the minimum number of nodes and is sized on the larger of the CPU need and the memory need. Pods are not packed perfectly, and unused space on a node is paid for.
- In pod mode you pay for the CPU and memory pods request, with no nodes and no minimum. Oracle's virtual nodes can't be sized from its price list, so that cloud is incomplete in this mode.
- One HTTP load balancer, sized as in the web app estimate (each request is about 2 KB and opens a new connection). Volumes are single-zone SSD, and Azure sells disks of 128 GB, so a volume is rounded up to whole disks.
- The cluster fee is the level with an uptime commitment: AKS Standard tier, EKS standard support, GKE regional cluster and OKE Enhanced cluster. Where a cloud has no level without a fee, the free choice uses the standard fee. Google's monthly free credit and Amazon's extended-support surcharge are not included.
- Container registry, logging and monitoring, NAT gateways, ingress add-ons, snapshots, spot nodes, reservations, support plans and taxes are not included.