TKE Serverless cluster is an out-of-the-box TKE service that allows you to deploy workloads without purchasing any nodes. It is fully compatible with native Kubernetes, allowing you to purchase and manage resources natively. This service is billed based on the actual amount of resources used by containers. In addition, it provides extended support for Tencent Cloud products, such as storage and network products, and can ensure the secure isolation of containers.
Containers are virtualization tools applied at the process level. With the ability to isolate and control system resources, containers restrict global resources access to processes in selected containers. A container image is a virtual machine snapshot and can be seen as the static form of a container. An image defines all files and dependencies required to run a container, ensuring consistency for running the container.
By containerization, all applications and their dependencies are packaged into an image, and then use the image to generate a resource-isolated environment to run the applications. This allows the applications to run independently in a consistent environment in a simple and efficient manner.
Kubernetes is an open-source Container Orchestration Engine (COE) inspired by a Google project called Borg. It is one of the most important components of the Cloud Native Computing Foundation (CNCF). Kubernetes provides production-level features such as application orchestration, container scheduling, service discovery, and autoscaling. For more information, see Kubernetes Documentation.
TKE Serverless cluster is a community-driven and out-of-the-box service, which supports the latest version of Kubernetes and native Kubernetes cluster management. It serves as a plug-in to provide extended support for Tencent Cloud products, such as storage, network, load balancing products.
TKE Serverless cluster is a fully-managed Kubernetes service, which means that you do not need to manage any computing nodes. It delivers computing resources by using Pods. It allows you to purchase, return, and manage cloud resources as in Kubernetes.
TKE Serverless cluster achieves 99.95% or higher availability based on the mature virtualization technology and network architecture of Tencent Cloud. Tencent Cloud ensures virtual isolation and network isolation between the TKE Serverless clusters of different users and allows users to configure network policies for a specific service by using services such as security groups and network ACL.
With the lightweight virtual technology developed by Tencent Cloud, you can create or delete a TKE instance in seconds to ensure higher efficiency. TKE Serverless cluster allows you to configure the native Horizontal Pod Autoscaler (HPA) of Kubernetes so that services can be automatically scaled based on actual loads.
The serverless architecture allows TKE Serverless clusters to provide higher resource utilization and lower Ops costs. The flexible and efficient autoscaling capability ensures that TKE instances only consumes the amount of resources required by the current load.
TKE Serverless cluster can be highly integrated with most Tencent Cloud services, including the storage products Cloud Block Storage (CBS), Cloud File Storage (CFS), and Cloud Object Storage (COS), TencentDB product family, and virtual private cloud (VPC) product family. With this capability, TKE Serverless cluster can provide solutions that meet the requirements of a wide range of businesses.
Three billing modes are available for TKE Serverless clusters: Reservation, pay-as-you-go, and spot mode. For more information, see Product Pricing.
|Feature||TKE General Cluster||TKE Serverless Cluster|
|Kubernetes||This feature is natively supported.||This feature is natively supported. Some features are not supported due to the lack of computing nodes. For more information, see Notes.|
|VPC||This feature is supported.|
|Computing nodes||You need to purchase and manage computing nodes such as Cloud Virtual Machine (CVM) and Bare Metal (BM) nodes on your own.||You do not need to purchase any computing nodes.|
|Management method||Native Kubernetes APIs and Kubectl are supported.|
|Clusters||Multiple clusters can be created and managed.|
|Namespaces||This feature is natively supported.|
|Workloads||This feature is natively supported.||Native Kubernetes workloads, except DaemonSet, are supported.|
|Service||This feature is natively supported. A CLB plug-in is integrated with TKE.|
|Storage||This feature is natively supported. Plug-ins such as CBS and CFS can be integrated with TKE.|
Running microservices with TKE Serverless clusters can free users from Ops of computing nodes. A service can be automatically scaled based on the actual load and use the necessary amount of resources to run applications, which reduces resource costs.
To run an offline computing task with a TKE Serverless cluster, you simply need to prepare a container image to quickly deploy workloads for the task. In addition, a TKE Serverless cluster bills only the actual amount of computing resources used during the execution of the task and stops billing when Pods are automatically released after the task ends.
TKE Serverless cluster can run online inference services by using CPU, GPU, and vGPU resources. The abundant resource specifications and the workloads that support autoscaling improve the operating efficiency and cost-effectiveness of the online inference services.