Help Center> Cloud Container Engine> User Guide> Old Console> Clusters> CCE Turbo Clusters and CCE Clusters
View PDF

CCE Turbo Clusters and CCE Clusters

Comparison Between CCE Turbo Clusters and CCE Clusters

The following table lists the differences between CCE Turbo clusters and CCE clusters:

Table 1 Cluster types

Dimensions

Sub-dimension

CCE Turbo Cluster

CCE Cluster

Cluster

Positioning

Next-generation container cluster for Cloud Native 2.0 with accelerated computing, networking, and scheduling

Standard cluster for common commercial use

Node type

Hybrid deployment of VMs and bare-metal servers

Hybrid deployment of VMs and bare-metal servers

Supported models

Models based on the QingTian architecture with software-hardware synergy

General-purpose models

Network

Network model

Cloud Native Network 2.0: applies to large-scale and high-performance scenarios.

Networking scale: 2000 nodes

Cloud-native network 1.0 for scenarios that do not require high performance or involve large-scale deployment.

  • Tunnel network model
  • VPC network model

Network performance

The VPC network and container network are flattened into one, achieving zero performance loss.

The VPC network is overlaid with the container network, causing certain performance loss.

Container network isolation

Pods can be directly associated with security groups to configure isolation policies for resources inside and outside a cluster.
  • Tunnel network model: Network isolation policies are supported for intra-cluster communication (by configuring network policies).
  • VPC network model: Isolation is not supported.

Security

Isolation
  • Bare-metal server: You can select secure containers for VM-level isolation.
  • VM: Common containers are deployed.

Common containers are deployed and isolated by Cgroups.

Performance Description on Batch Creating Pods in a CCE Turbo Cluster

Pods in a CCE Turbo cluster request elastic network interfaces (ENIs) or sub-ENIs from VPC. Currently, pods are bound with ENIs or sub-ENIs after pod scheduling is complete. The pod creation speed is constrained by how fast network interfaces are created and bound. The following table describes the constraints.

Table 2 Network interface creation duration

Node Type

Network Interface Type

Binding Network Interface to Node

Network Interface Availability

Concurrency Control

Default Node Pre-binding

ECS

Sub-ENI

Specify the ENI of the node to create a sub-ENI.

Within one second

Tenant-level: 600/minute

No pre-binding

BMS

ENI

Bind an ENI to a node.

20s-30s

Node-level: 3 concurrently

Pre-binding based on the threshold ratio 0.3:0.6

Creating Pods on ECS Nodes (Using Sub-ENIs)

  • If no pre-bound ENI is available on the node to which the pod is scheduled, the API for creating a sub-ENI is called to create a sub-ENI on an ENI of the node and allocate the sub-ENI to the pod.
  • If a pre-bound ENI is available on the node to which the pod is scheduled, the earliest unused sub-ENI is allocated to the pod.
  • Limited by the concurrent creation speed of sub-ENIs, a maximum of 600 pods can be created per minute without pre-binding. If a larger-scale creation is required, you can configure pre-binding for sub-ENIs.

    Pre-binding consumes subnet IP addresses and constraints the number of pods that can run in the cluster. Plan the pre-binding ratio accordingly.

Creating Pods on BMS Nodes (Using Sub-ENIs)

  • If no prebound ENI is available on the node to which the pod is scheduled, the API for binding an ENI to the node is called to bind and allocate an ENI to the pod. It takes about 20 to 30 seconds to bind an ENI to a BMS node.
  • If a pre-bound ENI is available on the node to which the pod is scheduled, the earliest unused ENI is allocated to the pod.
  • Due to the limited speed of binding ENIs to a BMS node, when no pre-binding is performed, 3 pods will be started in 20 seconds on a node. Therefore, 30% to 60% of all ENIs of a BMS node are pre-bound by default.

ENI pre-binding policies (checked every 2 minutes):

  • If the number of pre-bound ENIs on a node plus the number of ENIs currently used by pods is less than the number of ENIs at the low threshold (node ENI quota multiplied by the low threshold ratio), some ENIs will be prebound to make the numbers equal.
  • If the number of pre-bound ENIs on a node plus the number of ENIs currently used by pods is larger than the number of ENIs at the high threshold (node ENI quota multiplied by the high threshold ratio), some ENIs will be released to make the numbers equal.
Parent topic: Clusters