Resource planning

Cloud resource costs can grow significantly at scale. Different resources have different pricing dimensions and tiers, so understanding pricing principles and planning ahead helps you save money. Effective resource planning aligns resources with business and cost requirements, balancing on-demand supply and provisioned capacity to support growth, handle failures, ensure high availability, and manage provisioning times.

Plan resources for the business roadmap

A predictable business roadmap helps you plan and reserve resources. You can then factor these resources into your cost estimates. Common business evolution scenarios include:

1. Deploying new services on the cloud, which requires resource planning for new applications and systems.

2. Adding new computing power to support the growth of existing services.

3. Migrating existing IDC systems to the cloud.

4. Adding new edge devices and security products for cloud network transformations in a hybrid cloud environment.

5. Increasing data storage and data transfer bandwidth.

Before migrating to the cloud, plan and design your cloud resource capacity to optimize costs. You can assess and plan capacity from two perspectives: permanent resources and elastic resources.

A permanent resource handles online services and long-running jobs. Plan capacity for these resources based on your actual workload levels. For high-reliability scenarios, set permanent capacity to match peak workloads. For less critical scenarios, match average workloads.

An elastic resource handles sudden traffic spikes and one-off tasks, such as sales promotions, course registrations, unexpected traffic bursts, and temporary tasks.

Survey the business plans of various departments to identify the appropriate resource specifications and quantities, while leaving a buffer. Create resource plans on a monthly, quarterly, or fiscal-year basis, and forecast future usage based on current consumption.

Plan Alibaba Cloud resource regions

The cost of Alibaba Cloud products and services varies by location, depending on demand and local infrastructure costs. For a complete list of regions, see Regions and Availability Zones.

Understand the characteristics of Alibaba Cloud infrastructure and product availability in different regions. Consider the following aspects:

1. The coverage of cloud products in different regions.

2. The pricing information on the cloud product details page or purchase page.

3. The launch schedule for cloud products in each region.

Plan computing resource supply

An instance is the smallest unit that provides computing services for your business. Different instance types offer different computing capabilities. The instance family documentation describes all available ECS instance families, including their features, available types, and applicable use cases.

The Resource Planning tool in Alibaba Cloud's Resource Manager provides cloud planning capabilities. You can use this tool to check if your resource supply is healthy. It combines factors like sufficiency, replenishment capability, and popularity to reflect the real-time supply health of a specific instance type, which helps you plan your resource procurement. View Resource Supply Health Score and reserve relevant resources:

• Resource Supply Planning allows you to view the supply of cloud resources, create a resource reservation plan, and implement it.

• Resource Specification Recommendation suggests the most suitable compute resource specifications and scale to meet your computing power needs for specific business scenarios.

Two best-practice approaches are available:

1. Recommending ECS resources based on physical server specifications. This method is suitable for scenarios where you are migrating on-premises IDC servers to the cloud. Based on your on-premises IDC server configuration information (including CPU model, number of CPU chips, and memory size), the tool recommends a matching Alibaba Cloud ECS resource plan to guide your capacity planning.

2. Recommending ECS resources based on total computing power. This method recommends a matching Alibaba Cloud ECS resource plan as a reference for your future resource procurement. It is suitable when you need to plan computing power for new business based on the usage of existing services.

The resource plan provides a cost estimate to help you with financial planning and resource provisioning and management. Alibaba Cloud also provides best practices for selecting an instance family based on typical application scenarios.

Plan resource billing methods

Choosing the right billing method is the most direct way to optimize costs. For stable, long-term workloads, use lower-cost options like Savings Plans and reserved instances. Start with a lower configuration, evaluate the workload, then upgrade if needed or release underutilized resources. For stateful workloads with a defined lifecycle, use flexible pay-as-you-go billing, such as for temporary testing or Auto Scaling. For stateless and fault-tolerant workloads, use low-cost preemptible instances.

Combine different resource types to save costs. For example, using Savings Plans with pay-as-you-go instances significantly reduces expenses compared to pay-as-you-go alone. Consider a daily peak in online classes from 5 PM to 11 PM that requires many resources: use Savings Plans for the off-peak load, and Auto Scaling to provision pay-as-you-go instances during peak hours. For more information, see the guide on using pay-as-you-go instances as an elastic resource group in Auto Scaling.

For stateless workloads with low computing power requirements, combine pay-as-you-go instances with preemptible instances. Use pay-as-you-go for baseline computing needs and add preemptible instances to lower costs. For details, see the Elastic Supply best practices.

Plan quotas

Public cloud resources are subject to quotas. You can View Quotas and Rights to see the quotas and privileges available to your account. When planning, check whether your current quotas meet your business resource needs.

• ECS instance quotas are divided into total quota, reserved quota, and guaranteed quota based on supply certainty. The total quota is the maximum number of instances you can run, while the guaranteed quota is the minimum supply guaranteed for that instance. The system periodically adjusts your instance quota based on your usage to ensure that the automatic growth of your total quota can meet your daily resource usage growth. If the total quota does not meet your requirements, you can submit a request to increase it.

• Increase quota limits for resources such as images, cloud disks, and security groups.

• Plan for special privileges, including real-time downgrades, image copying, and image exporting.

Elasticity Assurance

Use Elasticity Assurance to secure resources for periodic, short-term, business-critical workloads. This is ideal when you need resources for a fixed period (daily, weekly, or monthly) but your overall usage is too low to justify a reserved instance.

With Elasticity Assurance, you pay a small fee to guarantee resource availability for a fixed period (1 month to 5 years). You set attributes like the availability zone and instance type, and the system reserves matching resources in a private pool. When you create pay-as-you-go instances, you can use capacity from this private pool to guarantee availability.

For example, a financial SaaS provider might need a large number of resources for reconciliation at the beginning of each month, or a rendering company might need to process a certain volume of rendering tasks at the beginning of each week.

Plan tags

Use tags to group, search for, and manage resources. Tags also support cost management and allocation.

When creating tags, plan them according to the following design principles:

• Use the collectively exhaustive principle to plan all your tags at once, and prioritize planning your tag keys. All resource objects must be bound to a planned tag key and its corresponding tag value. This principle is essential for using tags for access control, cost tracking, automated operations, and group searches.

• Use the mutually exclusive principle to avoid using two or more tag keys for the same attribute. Use the limited values principle to limit tag values to a core set. This simplifies resource management, access control, automated operations, and billing processes.

• Consider the consequences of future changes. When planning tags, think about the impact of adding or removing tag values later to improve flexibility. Modifying a tag can affect tag-based access control, automated operations, or related billing reports.

• Simplify your design. Use a fixed set of tag keys when planning your tags. This reduces operational errors caused by having too many tag keys.

You can refer to these tag design examples when planning your tags.

Plan storage usage

Alibaba Cloud offers the Storage Capacity Unit (SCU) and storage resource plan to help you manage your storage resource usage.

A Storage Capacity Unit (SCU) is a prepaid storage capacity plan that offsets pay-as-you-go bills for various cloud storage products. Unlike single-product plans, an SCU works with multiple cloud products, offering both cost-effectiveness and flexibility.

A storage resource plan lets you use the same amount of resources at a lower price. Use the following methods to plan the resource plan type and quantity:

• Choose a resource plan type based on your use case.

• Choose a resource plan type based on the billable items in your bill.

• Choose a resource plan specification based on the usage in your bill.

• Troubleshoot why you are still being charged hourly after purchasing a plan.

You can view the usage and remaining amount of your Storage Capacity Unit and storage resource plan in the Resource Instance Management Tool.

Plan traffic usage

Plan your public traffic and traffic resource usage. Purchase traffic resource plans to offset outbound public traffic charges. Use a NAT Gateway to manage inbound and outbound traffic and share bandwidth.

ECS types for container clusters

For dedicated container clusters, plan the instance types of the host machines. Larger ECS instance types are recommended.

Advantages of larger ECS instance types:

• Higher network bandwidth, leading to better resource utilization for high-bandwidth applications.

• Increased likelihood of containers communicating within the same ECS instance, which reduces network traffic.

• More efficient image pulling. An image only needs to be pulled once to be used by multiple containers. With smaller ECS instance types, images are pulled more frequently. If you need to scale the cluster with ECS Auto Scaling, it will take more time, delaying the scaling response.

Using many small-spec ECS instances when creating a Kubernetes cluster has the following disadvantages:

• The network resources of a small-spec worker ECS are limited.

• If a container occupies most of a small-spec ECS, the remaining resources on that instance may be insufficient for creating new containers or recovering failed ones. This can lead to resource waste when you use many small-spec ECS instances.

Marketplace services and billing

Alibaba Cloud Marketplace provides both Alibaba Cloud services and third-party vendor services. Billing structures vary by category and include free, pay-as-you-go, one-time purchase fees, or managed products/services.

Plan resources for cost optimization

Consider the following methods for resource reduction planning:

1. Metering and pricing vary by product and often have different pricing tiers based on the resource's location, size, or capacity. Infrastructure-related costs are relatively small.

2. Move shared resources to a shared resource group to maximize utilization and allocate costs across different business departments.

3. In your cost analysis, identify high-cost and underutilized resources to begin your resource reduction plan.

4. Break down cost optimization goals by resource group and billing unit to set targeted objectives.

5. Follow cost optimization best practices in your design. Optimize your architecture to reduce resource scale and costs.

6. Prioritize cost optimization. Planning resources before provisioning and managing them helps control costs.

7. Use Alibaba Cloud Savings Plans.

Combine these methods to achieve your cost optimization goals.