{"id":20607,"date":"2021-03-25T06:33:18","date_gmt":"2021-03-25T06:33:18","guid":{"rendered":"http:\/\/www.devopsschool.com\/blog\/?p=20607"},"modified":"2025-01-23T12:45:31","modified_gmt":"2025-01-23T12:45:31","slug":"kubernetes-as-a-service-with-comparison-of-eks-vs-aks-vs-gke","status":"publish","type":"post","link":"https:\/\/www.devopsschool.com\/blog\/kubernetes-as-a-service-with-comparison-of-eks-vs-aks-vs-gke\/","title":{"rendered":"Kubernetes as a service with comparison of EKS vs AKS vs GKE"},"content":{"rendered":"\n

This blog is still in progress. I am still working on it. Howee]ver, I have consolidated all the comparison available on the internet at one place in image format. Please refer it.<\/p>\n\n\n\n

Feature Comparison<\/h2>\n\n\n\n

In this section, we compare the key features of the three providers. Following this table, we\u2019ll provide a deeper analysis of each feature.<\/p>\n\n\n\n

 <\/th>Amazon EKS<\/strong><\/th>Microsoft AKS<\/strong><\/th>Google GKE<\/strong><\/th><\/tr><\/thead>
Supported Kubernetes version(s)<\/strong><\/td>1.18
1.17
1.16
1.15<\/td>		1.20
1.19
1.18
1.17<\/td>		1.17
1.16
1.15
1.14<\/td><\/tr>
Service Launch Date<\/strong><\/td>June 2018<\/td>June 2018<\/td>August 2015<\/td><\/tr>
CNCF Kubernetes Conformance<\/a><\/strong><\/td>Yes<\/td>Yes<\/td>Yes<\/td><\/tr>
Control-plane Upgrade<\/strong><\/td>Manual<\/a>
User must also manually update the system services that run on nodes (e.g., kube-proxy, coredns, AWS VPC CNI)<\/td>		Manual<\/a>
All system components update with cluster upgraded<\/td>		Automatic<\/a> (default) or Manual<\/a><\/td><\/tr>
Node Upgrade<\/strong><\/td>Manual<\/a>
EKS will drain and replace nodes<\/td>		Automatic or manual<\/a>
AKS will drain and replace nodes<\/td>		Automatic (default)
GKE drains and replaces nodes<\/td><\/tr>
Node OS<\/strong><\/td>Linux:
Amazon Linux 2<\/a> (default)
Ubuntu (Partner AMI)
Bottlerocket<\/a>
Windows:
Windows Server 2019<\/a><\/td>		Linux:
Ubuntu<\/a>
Windows:
Windows Server 2019<\/a><\/td>		Linux:
Container-Optimized OS (COS) (default)
Ubuntu
Windows:
Windows Server 2019<\/a>
Windows Server version 1909<\/a><\/td><\/tr>
Container Runtime<\/strong><\/td>Docker (default)
Containers (through Bottlerocket)<\/a><\/td>		Docker (default)
Containerd<\/a><\/td>		Docker (default)
Containerd
GVisor<\/a><\/td><\/tr>
High Availability Cluster<\/strong><\/td>Control plane is deployed across multiple Availability Zones<\/a> (default)<\/td>Control plane components will be spread between the number of zones defined by the Admin<\/a><\/td>Zonal Clusters:
Single Control Plane<\/a>
Regional Clusters:
Three Kubernetes control planes quorum<\/td><\/tr>
Control Plane SLA<\/strong><\/td>99.95%<\/a><\/td>99.95%<\/a><\/td>99.95%<\/a><\/td><\/tr>
Pricing<\/strong><\/td>$0.10\/hour (USD)<\/a> per cluster + standard costs of EC2 instances and other resources<\/td>Pay-as-you-go<\/a>: Standard costs of node VMs and other resources<\/td>$0.10\/hour (USD)<\/a> per cluster + standard costs of GCE machines and other resources<\/td><\/tr>
GPU support<\/strong><\/td>Yes (NVIDIA)<\/a><\/td>Yes (NVIDIA)<\/a><\/td>Yes (NVIDIA)<\/a><\/td><\/tr>
<\/td>Required install device plugin<\/td>Required install device plugin<\/td>Required install device plugin
Compute Engine A2 VMs<\/td><\/tr>
RBAC<\/strong><\/td>Yes<\/a><\/td>Yes<\/a><\/td>Yes<\/a><\/td><\/tr>
Control Plane: Log Collection<\/strong><\/td>Optional
Default: Off
Logs are sent to AWS CloudWatch<\/td>		Optional
Default: Off
Logs are sent to Azure Monitor<\/td>		Optional
Default: Off
Logs are sent to Stackdriver<\/td><\/tr>
Container Performance Metrics<\/strong><\/td>Optional
Default: Off
Metrics are sent to AWS CloudWatch Container Insights<\/a><\/td>		Optional
Default: Off
Metrics are sent to Azure Monitor<\/a><\/td>		Optional
Default: Off
Metrics are sent to Stackdriver<\/a><\/td><\/tr>
Node Health Monitoring<\/strong><\/td>No Kubernetes-aware support<\/a>; if node instance fails, the AWS autoscaling group of the node pool will replace it<\/td>Auto repair<\/a> is now available. Node status monitoring is available<\/a>. Use autoscaling rules<\/a> to shift workloads.<\/td>Node auto-repair enabled<\/a> by default<\/td><\/tr>
Serverless Compute<\/strong><\/td>AWS Fargate<\/a><\/td>Azure Container Instances<\/td>Cloud Run for Anthos<\/td><\/tr>
On-Prem Services<\/strong><\/td>Via AWS Outposts<\/a><\/td>Yes<\/td>Via Anthos GKE On Prem<\/a> through Google\u2019s Connect<\/a> service for multicluster management, in a vSphere 6.5 or 6.7 environment<\/a><\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

Strengths and Weaknesses of AKS, EKS, GKE<\/strong><\/h2>\n\n\n\n

In alphabetical order, here is a quick summary of AKS, EKS and GKE. It is not an exhaustive list and could be debated as some strengths or weaknesses you might view as a strength for your individual application.<\/p>\n\n\n\n

AKS Strengths<\/strong><\/h3>\n\n\n\n