Elastic TKE

Last updated: 2020-02-14 15:46:36



Elastic TKE (Elastic Kubernetes Service,EKS) is a service model launched by Tencent Cloud and TKE that can deploy workloads without requiring users to purchase nodes. EKS is fully compatible with native Kubernetes,. You can use native to purchase and manage resources, and charge according to the actual amount of resources used by the container. EKS also extends support for Tencent Cloud's storage and network products, while ensuring that user containers are securely isolated and used right out of the box.

Relevant Concepts

Container and Image

Container is a lightweight virtualization technology at the operating system level, which isolates and controls system resources so that the original global resources can only be used by processes in the container. Image is similar to a lightweight virtual machine snapshot and can also be understood as a static form of a container. Image defines all the files and dependencies that the container runs, which ensures the consistency of the container operation.
Container technology simply and efficiently realizes the independence and environmental consistency of the application by packaging the application and its dependencies into Image, and then using Image to generate a resource-isolated environment to run the program.


Kubernetes is Google's open source container orchestration and scheduling engine based on Borg, and it is one of the most important components of CNCF (Cloud Native Computing Foundation). Provides production-level application choreography, container scheduling, service discovery, automatic capacity expansion and other functions. For more information, please see Kubernetes official documentation .


Native supports

EKS follows the community and supports the latest Kubernetes version and native's Kubernetes cluster management style. Expand to support Tencent Cloud series of products in the form of plug-ins, such as storage, network, Cloud Load Balancer and other services, right out of the box.

No server

EKS is a fully managed Kubernetes service, which means that users do not have to manage any compute nodes. EKS delivers computing resources in the form of Pod, which allows users to purchase, return and manage cloud resources using Kubernetes native.

Security and Reliability

Based on Tencent Cloud's mature virtualization technology and network architecture, it provides 99.95% service availability. Tencent Cloud ensures virtualization isolation and network isolation of EKS clusters among users, and supports users to configure network policies for specific services through security groups, network ACL and other products.

Second stretch

Through Tencent Cloud Tencent's 's lightweight virtualization technology to ensure faster resource creation efficiency, users can create or delete TKE in seconds. EKS supports setting Kubernetes native HPA, which allows the service to scale automatically according to the actual load.

Reduce cost

The serverless form determines that EKS can bring users higher resource utilization and lower OPS cost. Auto Scaling's flexible and efficient ability ensures that TKE will only use the resources required by the current load.

Service integration

EKS can be highly integrated with most of Tencent Cloud's businesses, such as storage products HDD cloud disk CBS, file storage CFS and Cloud Object Storage COS, cloud database TencentDB series products, VPC VPC series products, etc., providing solutions to meet various business needs.

Use Limits

For more information, please refer to Purchase restriction .


EKS is a fully hosted serverless Kubernetes service. Like a TKE hosting cluster, it does not charge cost for managed resources such as Master,Etcd. The workload running in the elastic cluster adopts Postpaid's Pay as you go mode. Cost is calculated according to the actual amount of resources allocated according to the usage time. For more information, please refer to Pricing .

Comparison with TKE

CharacteristicsTKE (TKE)Elastic TKE (EKS)
KubernetesNative supportsNative supports it. There is a default function caused by no computing node. For more information, please seeDefault Kubernetes function
Virtual Private Cloudsupported
Compute nodeUsers are required to purchase and manage computing nodes such as CVM, bare metal independently.No need to purchase nodes
Management modeSupport native Kubernetes API, Kubectl and other methods
ClusterSupport to create and manage multiple clusters
NamespaceNative supports
WorkloadNative supportsSupport other native workload types except DaemonSet
ServiceNative supports the integration of CLB plug-ins
StorageNative supports and integrates plug-ins such as CBS, CFS, etc.

Use Cases

Microservice scene

Use EKS to run microservice to exempt users from OPS work on the computing node. The service can scale automatically according to the load, use the most reasonable amount of resources to host applications, and reduce the cost of using resources.

Offline computing scenario

Using EKS to run offline calculation of job, you only need to prepare the container Image to quickly deploy job load. In addition, EKS only charges cost, who is calculated by job's real running time, and job ends the billing after the automatic release of Pod.

Online reasoning scenario

EKS supports the use of CPU, GPU and vGPU to run online reasoning services. Rich resource specifications and Auto Scaling's load make running services more efficient and economical.

  • Storage: if you need to use HDD cloud disk or file storage as a container for persistent storage, you can use the HDD cloud disk CBS and File storage CFS .
  • Network:
  • To create and manage your VPC, such as creating VPC and subnets, establishing peering connections, using NAT gateways, configuring Route table, configuring security policies, etc., you can use VPC VPC .
  • To manage the private and public network Access configuration of the service, you can use the Cloud Load Balance CLB .
  • For API:, if you need to use Tencent Cloud API to use Access Tencent Cloud's products and services, please refer to Tencent Cloud API Documentation .