Tencent Cloud

Recent Pages

Node Pool Overview

Last updated: 2024-12-26 17:32:51

To help you efficiently manage nodes in a Kubernetes cluster, Tencent Kubernetes Engine (TKE) introduced the concept of node pools. A node pool is a logical set of nodes with similar attributes. Multiple node pools with different configurations and types can be created in a cluster. The configurations of a node pool include the resource attributes and Kubernetes attributes of nodes, such as billing mode, specifications, availability zone, tag, and taints. The basic features of node pools enable you to conveniently and quickly create, manage, and terminate nodes, thereby achieving more flexible and efficient resource management.
Use Cases
In business scenarios requiring large-scale clusters, it is recommended to use node pools for node management to improve the usability of such clusters. The following are applications of node pools in various large-scale cluster management scenarios:
1. Batch management: Through node selection based on resource and performance calculations, services are deployed on a batch of nodes with the same configurations.
2. Auto scaling: All nodes in a node pool bear the same business. During peak or off-peak times, machines can be automatically scaled in or out.
3. Specified scheduling: By using node pool tags, you can quickly specify business scheduling rules.
4. Physical isolation: Different node pools in the same cluster are used to distinguish business resources and test resources.
Product Architecture
Generally, all nodes in a node pool share the following attributes:
Billing mode: currently supports pay-as-you-go, spot instance, and monthly subscription.
Operating system: the type of the operating system on the node.
Hardware specifications: include CPU, memory, and GPU specifications.
Startup parameters: the startup parameters for Kubernetes components of the node.
Startup script: the custom startup script for the node.
Tag and taint settings: the Kubernetes Label and Taint settings for the node.
﻿
Documentation
For details, see Cluster Autoscaler.