跳至內容

EKS or fargate which one to use?

0

I understand that it depends on case to case when to use kubernetes or when to use fargate? But, actually what is the exact functionalities needed which suggest to use fargate or kubernetes? If I have to suggest a customer which one to use at what business use case, how should I define it?

已提問 3 個月前檢視次數 122 次

2 個答案
0

Hello Team, Can anybody suggest on the query and validate the response please.

已回答 3 個月前

  • The answer provided by the re:Post agents is correct.

    PS: Regarding the differences between EKS with and without Fargate, you can find excellent detailed information in the official AWS documentation, AWS blogs, and existing re:Post discussions. I would highly recommend checking those resources first—it is often the quickest way to find answers to common questions. If you are looking to build a strong technical foundation, the AWS Skill Builder and official training paths are also fantastic resources

0

I can help clarify the relationship between Amazon EKS and AWS Fargate, as there seems to be some confusion in your question.

Amazon EKS and AWS Fargate are not mutually exclusive options - they actually work together. Amazon EKS is a managed Kubernetes service, while AWS Fargate is a compute option that can be used with Amazon EKS to run your containers.

When using Amazon EKS, you have different compute options for running your workloads:

1. Amazon EKS with EC2 instances - You manage the underlying infrastructure (virtual machines) that run your containers. This gives you full control over instance types, purchasing options (including Spot instances and GPU instances), and allows you to leverage EC2 discounts.

2. Amazon EKS with AWS Fargate - AWS manages the underlying infrastructure for you. You don't provision, configure, or scale virtual machines. Fargate provides on-demand, right-sized compute capacity and runs each pod in its own isolated compute environment with dedicated resources.

When to recommend Fargate with EKS:

  • Customer wants to eliminate infrastructure management overhead
  • Workloads don't require specific EC2 instance types, GPUs, or Windows containers
  • Application can run in private subnets with NAT gateway access
  • Customer prioritizes operational simplicity over cost optimization through reserved instances or spot pricing
  • Workload isolation at the kernel level is important

When to recommend EKS with EC2:

  • Customer needs GPU instances, Inferentia chips, or Arm processors
  • Workloads require specific EC2 instance types or purchasing options
  • Customer wants to leverage existing EC2 discounts or reserved instances
  • Need to run DaemonSets or privileged containers
  • Require full Kubernetes conformance with all upstream primitives
  • Cost optimization through instance selection is a priority

Important note: Amazon EKS Auto Mode is now the recommended approach moving forward. It provides the ease of use similar to Fargate while offering full Kubernetes conformance, support for all EC2 runtime purchase options (including GPU and Spot instances), and the ability to leverage negotiated EC2 discounts - capabilities not available with Fargate.

For business use cases, consider recommending Fargate when operational simplicity is the top priority and the workload fits within Fargate's constraints. Recommend EKS with EC2 (or EKS Auto Mode) when customers need flexibility, cost optimization options, or specific hardware requirements.
Sources
Migrate from EKS Fargate to EKS Auto Mode - Amazon EKS
Containers services on AWS - Containers on AWS
Simplify compute management with AWS Fargate - Amazon EKS
EKS Fargate Support - Eksctl User Guide

已回答 3 個月前

專家

已審閱 3 個月前

您尚未登入。 登入 去張貼答案。

一個好的回答可以清楚地回答問題並提供建設性的意見回饋,同時有助於提問者的專業成長。