Help Center/ Cloud-Native Application Modernization/ Alauda Container Platform/ Resource and Cost Planning
Updated on 2024-06-19 GMT+08:00
View PDF
Share

Resource and Cost Planning

Table 1 Resource and cost planning

Cloud Service

Usage

Region

Specifications

Quantity

Billing Mode

Billing Period

Billing Period Unit

Reference Price

ECS

For a global cluster, which carries all ACP platform functions (such as platform components, UI, Kubernetes master node resources, logs, and monitoring)

AP-Bangkok

Specifications: x86 computing | General computing-plus | c6s.2xlarge.2 | 8 vCPUs | 16 GiB; Image: CentOS | CentOS 7.9 64bit; System disk: General-purpose SSD | 150 GiB; Data disk 1: General-purpose SSD | 500 GiB; Data disk 2: General-purpose SSD | 100 GiB

3

Yearly/Monthly

1

Year

USD7,412.25

ECS

Master node in a service cluster (for Kubernetes master node resources, logs, and monitoring)

AP-Bangkok

Specifications: x86 computing | General computing-plus | c6s.xlarge.2 | 4 vCPUs | 8 GiB; Image: CentOS | CentOS 7.9 64bit; System disk: General-purpose SSD | 100 GiB; Data disk 1: General-purpose SSD | 100 GiB

3

Yearly/Monthly

1

Year

USD3,132.60

ECS

Prometheus monitoring node in a service cluster

AP-Bangkok

Specifications: x86 computing | General computing-plus | c6s.xlarge.2 | 4 vCPUs | 8 GiB; Image: CentOS | CentOS 7.9 64bit; System disk: General-purpose SSD | 100 GiB; Data disk 1: General-purpose SSD | 100 GiB

1

Yearly/Monthly

1

Year

USD1,044.20

ECS

Slave node in a service cluster (The price listed is for a single server. The actual server quantity required will depend on the specific requirements.)

AP-Bangkok

Specifications: x86 computing | General computing-plus | c6s.2xlarge.2 | 8 vCPUs | 16 GiB; Image: CentOS | CentOS 7.9 64bit; System disk: General-purpose SSD | 100 GiB; Data disk 1: General-purpose SSD | 100 GiB

1

Yearly/Monthly

1

Year

USD1,871.25

Virtual IP address (VIP)

High availability (HA) in a Kubernetes cluster

AP-Bangkok

/

2

Yearly/Monthly

1

Year

USD0.00

Elastic IP (EIP)

Platform access address

AP-Bangkok

Bandwidth price: Dedicated | Dynamic BGP | Billed by bandwidth | 5 Mbit/s

1

Yearly/Monthly

1

Year

USD243.00
  • The prices listed in the table are for reference purposes only. They may differ from the actual price, which is displayed on the Huawei Cloud console.
  • The table above shows the minimum recommended configurations for deploying ACP. Make sure to allocate resources according to specific service needs. If more information about the pricing of ACP is needed, contact your account manager.

Table 2 Hardware configuration (adjusted based on the service pressure)

Server Role

Master,

slave,

global,

log, and monitoring

Master node in a service cluster

Prometheus in a service cluster

Slave node in a service cluster

Number of Server

3

3 x 2 (two service clusters)

1 x 2

10 x 2

Server Function

Used to carry all functions of the platform

Master node in a service cluster

Monitoring node in service cluster

Slave node in a service cluster

Mandatory or Optional

Mandatory

Mandatory

Mandatory

Mandatory

Number of CPU

8

4

4

8

Memory Capacity

16 GiB

8 GiB

8 GiB

16 GiB

Available Space of the / Partition

150 GiB

50 GiB

50 GiB

50 GiB

/cpaas/data/

500 GiB for storing logs, separate block device

/

/

/

/cpaas/monitoring/

100 GiB for storing monitoring data

/

50 GiB

/

Available Space of /var/lib

50 GiB

50 GiB

50 GiB

50 GiB

Available Space of /opt

30 GiB

30 GiB

30 GiB

30 GiB

/var/lib/docker or /var/lib/containerd

100 GiB

100 GiB

100 GiB

100 GiB

/var/lib/docker or /var/lib/containerd

xfs

xfs

xfs

xfs

(Optional) Separate Block Device

/

100 GiB for topolvm

/

50 GiB x 2
Table 3 Hardware requirements

Hardware

Requirement

Model or Minimum Configuration

CPU

The dominant frequency must be at least 2.5 GHz and cannot be overcommitted at the IaaS layer. If requirements still cannot be met, you need to increase the number of CPUs. If Arm CPUs are used, you can increase the number of CPUs by 1.5x. It is advisable to increase the number of CPUs by 2x.

Intel 8255c

Memory

Overcommitment at the IaaS layer is not supported.

6-channel DDR4

Hard disk

The IOPS of a single block device is greater than 2,000, and the throughput is greater than 200 MB/s.

ssd

GPU

GPUs with driver 418.87.00 CUDA 10.1 have been fully tested.

Nvidia
Table 4 Network resource requirements

Resource

Mandatory or Optional

Quantity

Description

Certificate

Optional

1

If a certificate is not provided, the deployment script will generate one automatically. However, when a user accesses the platform UI through a browser, a security warning will appear because the certificate is not issued by a trusted certification authority.

Platform access address (external IP address)

Mandatory

1

Domain name or IP address. For details, see Platform Access Address in "Glossary" in Alauda Cloud Native Success Platform Installation Guide.

Global VIP

Mandatory

1

For details, see global VIP in "Glossary" in Alauda Cloud Native Success Platform Installation Guide.

Kubernetes API server VIP

Mandatory

Multiple

This resource is mandatory in the production environment and is used by kube-api of an HA Kubernetes cluster. Each HA Kubernetes cluster requires a VIP.

ALB VIP

Mandatory

Multiple

If customers require HA with ALB, this resource is necessary. Each service cluster's load balancer needs a VIP. (Note that it is the load balancer, not each ALB instance, that requires a VIP.)

ASM Istio gateway VIP

Optional

Multiple

For each service cluster where ASM Istio is deployed, if the HA Istio gateways are required, a VIP must be provided.

Private load balancer

Mandatory

1

This resource is essential in the production environment. Without it, HA requirements cannot be fulfilled. HA is achieved through load balancers like F5 load balancers. The load balancer is configured with both the Kubernetes API server VIP and the global VIP.

Public load balancer

Mandatory

1

This resource is essential in the production environment. Without it, HA requirements cannot be fulfilled. If the external networks are not distinguished from the internal networks, the private load balancer can be reused. The external address is configured on this load balancer.

More access addresses

Optional

Multiple

To use more IP addresses or domain names other than external address to access the global platform, prepare the domain names and IP addresses and add them in the advanced settings on the installation page during platform deployment.
Table 5 Network configuration requirements

Item

Requirement

Network rate

The bandwidth must be at least 1 Gbps. 10 Gbps is recommended. If the global platform and service cluster are deployed in different data centers or hybrid clouds, the network rate between them must be at least 100 Mbps. (1 Gbps will be better.) If there is no need to collect data such as service logs and audits from the service cluster, the required bandwidth can be reduced somewhat.

Network latency

The network latency should be 2 ms or lower. If the global platform and the service cluster are deployed in different data centers or in a hybrid cloud, ensure that the network latency between them is within 30 ms and does not exceed 100 ms.

Security and firewall

There is no firewall between servers on the global platform.

There is no firewall between servers in the service cluster.

It is recommended that no firewall be deployed between the service cluster and the platform. If a firewall is deployed between them, allow necessary ports to pass through the firewall by referring to Global platform four-layer forwarding rules and Compute cluster forwarding rules in "Network Requirements" in Alauda Cloud Native Success Platform Installation Guide.

Calico uses the IP in IP protocol. If the service cluster uses the Calico plug-in, the IP in IP protocol cannot be restricted.

IP address range

IP addresses in the 172.16 to 172.32 CIDR block cannot be used by the server hosting the platform. If any of these IP addresses are already being used, it cannot be altered. To work around this issue, you must modify the Docker configuration on each server and add the bip parameter to avoid using the CIDR block.

Agreement

If dual-stack networking is used, IPv6 must be supported.

Route

The server has a default route or a route pointing to 0.0.0.0.

Forwarding

All forwarding ports should be allowed.
Table 6 OSs and kernel versions

Architecture

CPU Model

Supported OS

Version Information

Remarks

Arm

Kunpeng 920

Kylin
  • Kylin Linux Advanced Server Release (Kylin V10) Kernel version:

4.19.90-11.ky10.aarch64

  • Kylin Linux Advanced Server Release (Kylin V10 SP1) Kernel version:

4.19.90-17.ky10.aarch64

  • Kylin Linux Advanced Server Release (Kylin V10 SP2) Kernel version:

4.19.90-24.4.v2101.ky10.aarch64

Virtualization is not supported.

openEuler

openEluer 20.03 SP3 Kernel version: 4.19.90-2112.8.0.0131.oe1.aarch64

openEluer 22.03 SP3 Kernel version: 5.10.0-60.18.0.50.oe2203.aarch64

None

x86

None

Kylin
  • Kylin Linux Advanced Server Release (Kylin V10) Kernel version:

4.19.90-11.ky10.x86_64

  • Kylin Linux Advanced Server Release (Kylin V10 SP1) Kernel version:

4.19.90-23.8.v2101.ky10.x86_64

  • Kylin Linux Advanced Server Release (Kylin V10 SP2) Kernel version:

4.19.90-24.4.v2101.ky10.x86_64

None

openEuler
  • 20.03 SP3 Kernel version: 4.19.90-2112.8.0.0131.oe1.x86_64
  • 22.03 SP3 Kernel version: 5.10.0-60.18.0.50.oe2203.x86_64

None

Ubuntu
  • Ubuntu 20.04 Kernel version: 5.4.0-124-generic
  • Ubuntu 22.04 Kernel version: 5.15.0-56-generic

None

Red Hat
  • Red Hat 7.8 Kernel version: 3.10.0-1127.el7.x86_64
  • Red Hat 8.0 Kernel version: 4.18.0-80.el8.x86_64
  • Red Hat 8.6 Kernel version: 4.18.0-372.9.1.el8.x86_64

None

CentOS

CentOS 7.6, 7.7, 7.8, and 7.9

Kernel version: 3.10.0-1160 and 3.10.0-1127

None