Resource Management

Alibaba Cloud Resource Management is a collection of enterprise IT governance products and services that includes Resource Directory, Resource Group, Resource Sharing, and Tag. Resource Directory builds your enterprise's organizational structure on the cloud, Resource Group and Tag manage cloud resources hierarchically, and Resource Sharing enables resource sharing among enterprise members. Combining these services to manage the entire resource lifecycle and consolidate costs is a best practice.

Manage resource directories and resource groups

Resource Directory lets you build a directory structure based on your enterprise's business or ecosystem and organize multiple accounts into a resource hierarchy. These organizational relationships enable centralized resource management and help meet governance requirements for finance, security, auditing, and compliance.

Set up a dedicated financial account to review and manage cloud costs at the enterprise level. This account pays for all other accounts within the organization and manages their bills, enabling centralized financial management. Use folders, accounts, or resource groups to allocate costs across different dimensions.

Resource Directory provides solutions for managing multiple accounts, tagging resources within a single account, and managing resource groups. These features support cost allocation by resource group, simplifying financial cost allocation. Select a resource management solution based on your security and auditing requirements.

Manage tags

Common scenarios for using tags include:

• Tag your resources, then use the Tag Console or an API to quickly find resources with specific tags. For more information, see Find resources using tags.

• Plan resource tags based on organizational or business dimensions, such as region, department, environment, or project, and use the cost analysis and split bill features of Alibaba Cloud to allocate costs. For more information, see Use tag-based cost allocation.

• Automate O&M by attaching different tags to different environments (such as production and staging), operating systems (such as Windows Server and Linux), or client platforms (such as iOS and Android). Then create and execute a template in Operation Orchestration Service (OOS) to perform batch automated O&M. For more information, see Use tags to automate O&M.

Manage resource instances

Resource Center provides a unified view of your resources across accounts, services, and regions. Search for resources by region, service, resource type, resource ID, resource name, or tag, then navigate directly to the corresponding service console for further operations.

Computing resources are typically divided into persistent and elastic resources. Use persistent resources to handle regular workloads, such as internal enterprise management platforms, online services, and long-running jobs. Estimate capacity needs based on workload levels, and use Cloud Monitor to obtain accurate utilization data and continuously track key metrics for optimization. Elastic resources handle unexpected traffic spikes and temporary tasks. With Auto Scaling, you can provision and automatically release resources on demand, preventing unnecessary costs from idle resources. Auto Scaling adjusts computing capacity, improves resource utilization, and reduces total cost of ownership while eliminating the need for manual resource adjustments. Use Auto Scaling to dynamically allocate resources based on business needs. For predictable changes, such as foreseeable peak hours, use time-based automatic scaling to deliver the right capacity when needed. You can also configure scheduled scale-out and scale-in activities for elastic resource groups based on workload analysis.

Using Alibaba Cloud APIs and automation to programmatically adjust cloud resource counts is a recommended practice. Automatically increase resources during peak demand to maintain performance, and reduce capacity when demand decreases to lower costs. Operation Orchestration Service (OOS) is an automated O&M service that manages and executes tasks. Define tasks, execution order, inputs, and outputs using templates, then execute them to automate operations. For example, if your business system experiences daily peak traffic periods requiring many instances, use OOS to schedule the startup and shutdown of ECS instances to save costs after peak hours.

Use elastic settings for database instances to adapt to peak business periods. Upgrade the instance type during peak periods to accommodate high traffic and ensure service stability, then reduce instance resources when workload decreases to save costs. The automatic scale-out and scale-in features of Database Autonomy Service (DAS) optimize database instance resource usage. For example, set a threshold for average CPU utilization of a database instance. When the threshold is reached, the instance type is automatically upgraded. When utilization reaches the scale-in threshold, cluster specifications are gradually scaled in until the cluster returns to its original specifications.

Reduce the cost of resource instances using resource plans and deduction plans. Resource plans include packages such as Storage Capacity Units (SCUs), storage plans, and PolarDB plans. After purchase, these resources become your assets and offset the usage of on-demand instances for specific resources, saving on billing costs. Use the Resource Management tool to view the summary and usage of resource plans and properly allocate deduction quotas. Note that different resource plans have different capacity types, and deduction rules for these capacity types may vary.

Manage resource quotas

Alibaba Cloud optimizes global platform resource allocation by setting quantifiable limits on resource counts (such as ECS instance types, images, cloud disks, and security groups) and operation counts (such as API call rates). A resource quota limit is the maximum quantity of a cloud resource you can use or the maximum number of operations you can perform. Use Quota Center to query quota limits and adjust quotas online based on business needs. Set quota alerts to reduce the business impact of insufficient resources. Alibaba Cloud provides a level of assurance for instance resource supply based on user rights. Use the quota management service of Enterprise Support to view and increase instance quotas.

Scale down and stop resources

Scale down idle resources and promptly stop unused resources, either manually or automatically, according to the resource lifecycle plan. When creating resources, use tags and groups to mark their lifecycle. For example, add tags such as 'machine for testing phase' or 'instance created for a 3-month task' to facilitate timely action.

Scale down resources based on business needs. If your system fully utilizes Auto Scaling, resources are automatically scaled down according to preset elastic scaling rules when business capacity decreases.

For resources without automatic scaling capabilities, promptly scale them down and stop them when business volume decreases or services are discontinued. Also scale down development and testing resources after a complete development cycle. For example, reduce the number of VPN client connections or downgrade ECS and RDS instance types.

View resource usage in monitoring tools and implement workload throughput monitoring or alerts to promptly address potential resource waste when throughput decreases. Use orchestration service and Cloud Monitor to automatically stop idle machines.

Prioritize serverless compute resources, such as Function Compute, to build serverless and event-driven architectures. Simply write and upload your code, and Function Compute prepares compute resources and runs your code elastically and reliably without infrastructure management. Serverless services provide automatic resource utilization optimization and automatic scaling. With serverless applications, you do not pay for over-provisioning.

Release resources

Promptly releasing unneeded resources saves cloud costs and reduces management costs and security risks from retaining unused resources. For example, release a temporary Elastic IP Address (EIP) used during development when it is no longer needed. Similarly, release temporary business traces used during system migration and cutover after the migration is complete. Use the pay-as-you-go billing method for temporary resources during resource planning so that releasing them stops charges immediately. To release a subscription resource mid-cycle without affecting your services, first convert it to pay-as-you-go, confirm there is no impact, then release it.

Manage data transfer and traffic

Manage the location of ongoing data transfers, their costs, and related business objectives. Monitor data transfer bandwidth and quality.

Release resources on unnecessary data links promptly. For example, during data migration to the cloud, temporary data links are often set up that may involve gateway products. After data migration is complete, dismantle these temporary data links and release their resources.

Incorporate data transmission into your security design. A proper security plan reduces unnecessary and risky data transfers. For traffic-billed data transfers, a good security plan mitigates data security risks and reduces unnecessary costs. Reduce unnecessary traffic and costs by enabling Cloud Firewall and using a protection whitelist to control inbound and outbound traffic of Elastic IP Addresses. Enable the Cloud Firewall access log to view client access activity. For clients that make excessive access attempts or operate beyond legitimate use, apply rate limiting promptly. For more information about security products, see the relevant sections in the Security Compliance pillar.

Manage storage resources

An architecture that separates storage and compute simplifies resource management. Manage storage plans based on products of different storage types. For example, for Object Storage Service (OSS), plan storage amounts for Standard, Infrequent Access, Archive, and Cold Archive storage classes. Define data retention and lifecycle policies to perform automatic storage class migration and deletion, reducing overall storage cost over the data lifecycle.

Track resource usage in workloads

Track the utilization of every resource in your workloads and adjust resources and instance types based on utilization changes. O&M engineers must monitor not only high utilization and high load but also low utilization and low load. For elastic and uncertain business scenarios, start with a lower resource configuration and scale up as needed.

Use the Cloud Monitor service to monitor metrics of various cloud resources and track their usage. Generate alerts based on baseline thresholds for each metric. Fine-tune alert thresholds based on changes in service traffic to reflect actual business usage and avoid over-responding.

Manage billing methods for resources

Optimize resource usage and costs by managing the billing methods of your resources. For more information, see Billing Method Optimization.