Auto Provisioning is a service to quickly deploy an instance cluster that consists of preemptible and pay-as-you-go instances. Auto Provisioning supports one-click deployment of instance clusters with specified billing methods, zones, and instance families. You can use auto provisioning groups to provide stable computing power, alleviate the instability caused by the reclaiming of preemptible instances, and eliminate the need to manually create instances.
Introduction to Auto Provisioning
Auto provisioning groups can create instance clusters to meet computing power needs based on configured instances types and scheduling policies. After an auto provisioning group is created, the auto provisioning group automatically selects instance types and creates an instance cluster. You do not need to calculate the costs of instances individually.
If you select Continuous Delivery and Maintain Capacity, the auto provisioning group automatically compares real-time and target capacities. When preemptible instances are reclaimed, the auto provisioning group selects instance types and creates new instances to maintain the target capacity of your business at peak demand and meet your computing power needs at the lowest cost.
Preemptible instances are low-cost computing resources that are subject to a protection period of one hour. After the protection period expires, preemptible instances and their resources may be reclaimed. You must pay attention to the availability of preemptible instances in the hour after the protection period. If the preemptible instances are unavailable, you must create new ones. As the number of preemptible instances increases, the time required to maintain them also increases. In this scenario, you can use auto provisioning groups to deploy instance clusters based on configured target capacities and scheduling policies.
Similar to preemptible instances, auto provisioning groups are applicable to stateless application scenarios such as scalable website services, image rendering, big data analytics, and parallel computing. Auto provisioning groups support flexible policy combinations to alleviate the impact of reclaimed preemptible instances. In addition, the delivery of instance clusters is convenient.
- Cross billing methods
Preemptible instances provide computing resources at low costs, but are restricted by the reclaim mechanism and available inventory in a region. Pay-as-you-go instances can be created and released at any time. These instances have a higher priority to consume resource inventory, but are much more costly compared to preemptible instances. Auto provisioning groups allow you to create both preemptible instances and pay-as-you-go instances. You can use both billing methods to reduce costs and meet your computing power needs.
- Cross zones
Deploying an instance cluster in the same zone reduces network latency between instances, while deploying an instance cluster across zones improves the disaster recovery capabilities of applications. Auto provisioning groups support the deployment of instance clusters across zones. You can set zone options as needed.
- Cross instance families
Auto provisioning groups allow you to specify alternative instance types across multiple instance families to provide a range of instance types to select from. Additionally, you can specify the weight and priority for each instance type to improve task scheduling while ensuring controllability.
- Flexible policy combinations
Auto provisioning groups can meet various dynamic business needs through combinations of target capacities and scaling policies. Auto provisioning groups allow you to set the target capacities of clusters, preemptible instances, and pay-as-you-go instances, and to specify scaling policies for preemptible instances and pay-as-you-go instances. Additionally, you can specify solutions to fulfill the target capacity of a cluster when preemptible and pay-as-you-go instances are insufficient.
- Complete cost control
Auto provisioning groups allow you to set a maximum price for all and individual instance types to ensure that costs remain within your expectations.
- Practical protection mechanism
Auto provisioning groups provide the shutdown option. You can enable this option when an auto provisioning group expires or when instances exceed the target capacity. Routine health checks are performed on instances in an auto provisioning group to ensure availability of the instances.
- Auto provisioning groups cannot schedule resources across regions.
- A maximum of 1,000 instances can be created in each auto provisioning group.
- Only a single launch template can be specified for each auto provisioning group. However, you can extend the launch template to implement more configurations. For more information, see Template configurations.