Billing Examples

Scenario 1

You purchased a pay-per-use CCE Autopilot cluster in AP-Bangkok for testing services on the cloud. The detailed operations are as follows:

1. At 15:30:00 on March 18, 2023, you purchased a CCE Autopilot cluster. The specifications are as follows:
   NOTE:

   For ease of description, assume that all the resources involved were created at the same time. There may be a time difference of several seconds. You can view the actual time on the console.

   • Billing mode: pay-per-use
   • SNAT: enabled
   • Add-ons: CoreDNS and Kubernetes Metrics Server. Two Deployments were automatically created for the add-ons:
     • Deployment for CoreDNS

       Pods: 2

       vCPUs: 1

       Memory: 2 GiB

       Ephemeral storage of a single pod: 30 GiB (free)

     • Deployment for Kubernetes Metrics Server

       Pods: 2

       vCPU: 1

       Memory: 2 GiB

       Ephemeral storage of a single pod: 30 GiB (free)

2. At 15:36:00 on March 18, 2023, you created a Deployment named nginx. The specifications are as follows:
   • Pods: 2 (one container in each pod)
   • Container image: nginx:latest
     • Container vCPU: 0.5 vCPUs
     • Container memory: 1 GiB
   • Create a LoadBalancer Service. The load balancer details are as follows:
     • Type: dedicated load balancer
     • AZ: single AZ
     • Specifications: network load balancing (TCP/UDP), small I (10 LCUs)
     • EIP type: dynamic BGP
     • EIP bandwidth: 5 Mbit/s, billed by traffic
   • Ephemeral storage of a single pod: 30 GiB (free)
3. At 09:36:00 on March 19, 2023, you decided to use the cluster for a long time and subscribed to four 1,000 vCPU-hours CPU monthly packages and seven 1000 GiB-hours memory monthly packages (expiration time of all packages: April 19, 2023, 23:59:59).
4. At 20:36:00 on April 5, 2023, you deleted the LoadBalancer Service for the Deployment.
5. At 21:06:00 on April 19, 2023, you deleted the cluster and related resources.

Figure 1 Operation timeline

Table 1, Table 2, and Table 3 list the incurred expenditures by resource type.

• In Table 1, the total expenditures of the cluster are 78.85 + 22.078 + 82.8365 = $183.7645 USD.
• In Table 2, the total expenditures of the pods are 0.0172 + 0.004 + 3.741+ 0.87 + 155.52 + 29.68 + 2.805 = 192.6372 USD.
• In Table 3, the total expenditures of the load balancer are 30.4015 + 1.08 = $31.4815 USD.

The total expenditures are $407.8832 USD.

NOTE:

*In pay-per-use billing, the resource usages are billed by the second and the amount due is truncated to the 2nd decimal place. This amount may vary depending on the actual bill.

Scenario 2

You deployed a large number of applications in a CCE Autopilot cluster in AP-Bangkok. It was estimated that 12,000 vCPUs and 48,000 GiB of memory were required for one year, so you purchased one Autopilot general-computing 12,000 vCPU-hours CPU yearly package and four Autopilot general-computing 12,000 GiB-hours memory yearly packages in advance. In the middle of the year, it was estimated that 36,000 vCPUs and 57,800 GiB of memory were required for new services. You wanted to know which billing mode was the most cost-effective for the new services.

NOTE:

*In pay-per-use billing, the amount due is truncated to the 2nd decimal place. This amount may vary depending on the actual bill.

• If the CCE Autopilot cluster is not required after the current year, the combination of pay-per-use billing and packages can help save costs. The following table compares the two billing modes in detail.

  Table 4 Billing mode comparison

  Billing Mode	Billing Rules	Expenditures*
  Pay-per-use	vCPU: Unit price × Number of vCPUs × Usage duration Memory: Unit price × Memory size × Usage duration	vCPU: 30,000 × 0.043 = $1,290 USD Memory: 57,800 × 0.005 = $289 USD Total: $1,579 USD
  Packages	3 x Autopilot general-computing 12,000 vCPU-hours CPU yearly package 5 x Autopilot general-computing 12,000 GiB-hours memory yearly package	vCPU: 3 × 388.8 = $1,166.4 USD Memory: 5 × 42.44 = $212.2 USD Total: $1,378.6 USD
  Pay-per-use and packages	vCPU: 3 × Autopilot general-computing 12,000 vCPU-hours CPU yearly package Memory: 4 × Autopilot general-computing 12,000 GiB-hours memory yearly package + 7 × Autopilot general-computing 1,000 GiB-hours memory monthly package + 800 GiB-hour pay-per-use memory	vCPU: 3 × 388.8 = $1,166.4 USD Memory: 4 × 42.44 + 7 × 4.24 + 800 × 0.005 = $203.44 USD Total: $1369.84 USD

• Assume that you want to use the CCE Autopilot cluster in the next year and about 80,000 to 100,000 vCPU-hours and 110,000 to 150,000 GiB-hours are required from the middle of the current year to the next year.
  • For vCPUs, you can purchase Autopilot general-computing 12,000 vCPU yearly packages in batches to prolong the validity period of packages and avoid waste. You can also flexibly adjust the package specifications based on the vCPU usage. For example, if the vCPU usage increases sharply in the next year, you can purchase a 120,000 vCPU-hours CPU yearly package.
  • For memory, you are advised to purchase an Autopilot general-computing 120,000 GiB-hours memory yearly package and configure a usage alert by referring to Configuring Package Usage Alerts so that you adjust the quota based on the subsequent usage when the package is about to run out.