Auto Scaling
Automated, low-cost computing resource management policy with planning capability, efficiency and high fault tolerance
Overview
Auto Scaling (AS) provides you a highly efficient management policy for computing resources. You can set the time to execute management policy regularly or create a real-time monitoring policy to manage the number of CVM instances and deploy the environment for the instances, to ensure that your business runs smoothly. AS automatically increases the number of CVM instances during demand surges to maintain good performance and decreases the number during lulls to reduce costs.
Benefits
Automatic
AS automatically creates and removes CVM instances in a real time and dynamic way based on the business load, ensuring that you are running the optimal number of instances without manual intervention.
Cost Saving
AS helps you maintain an optimal number of instances for variable business demands. When the demand rises, AS will automatically add new CVM instances rapidly, and when the demand drops, AS will automatically remove unnecessary instances accordingly. This improves device utilization and reduces the costs of deployment and instances.
Schedulable
AS allows you to set a schedule to plan the scaling activities to deal with regular changes in business load (e.g. scaling up at 21:00 every day).
Fault Tolerance
AS automatically checks instances' health. Once AS detects a faulty instance, it will automatically create a healthy instance to replace the faulty one. This ensures that your application is getting the computing capacity as you expect so that your business can run normally and smoothly.
Easy to Audit
When using auto scaling, the user can record the trigger conditions, time, involved instances and success/failure reasons related to each scaling event. Visualized tracking, querying interfaces and SMS notifications are provided to help you locate root causes as soon as possible for prompt solution.
Features
Alarm-triggered Scaling
If you want to adjust the business deployment based on CVM metrics, you can customize an alarm policy. When a metric (e.g. CPU utilization, memory utilization, outbound and inbound bandwidth on the private network, outbound and inbound bandwidth on the public network) reaches the threshold due to business load, the policy can automatically increase or decrease the number of CVM instances to deal with the changes in business load flexibly, improving device utilization and reducing costs of deployment and instances. The monitoring cycle is 1 minute.
Scheduled Scaling
When your business load changes are predictable (knowing the maximum number, minimum number and expected number of CVM instances), you can set a schedule to plan your scaling activities. You can use this feature to automatically increase or decrease the number of instances as scheduled or regularly. Daily/weekly/monthly repeated scaling activities are supported. This way you can meet the ever-changing business load easily, improve device utilization and reduce costs of deployment and instances.
Replacement of Unhealthy CVM
AS automatically checks CVM instances' health. Once AS detects a faulty CVM, it will automatically create a healthy instance to replace the faulty one. This ensures that your application is getting the computing capacity as you expect so that your business can run normally and smoothly.
Auto Load Balancing
CVM instances added via the alarm policy can be directly associated with the existing Cloud Load Balance (CLB). Then they will help to handle the traffic automatically distributed to them, improving the service availability.
Action: Added CVMs are automatically registered with CLB
Weight settings: The CLB weight for the added CVMs is set to 10 by default
CLB policy: Follows the policy defined in the CLB.
Learn more about CLB >
AS Activity Record
If you want to adjust the business deployment based on CVM metrics, you can customize an alarm policy. When a metric (e.g. CPU utilization, memory utilization, outbound and inbound bandwidth on the private network, outbound and inbound bandwidth on the public network) reaches the threshold due to business load, the policy can automatically increase or decrease the number of CVM instances to deal with the changes in business load flexibly, improving device utilization and reducing costs of deployment and instances. The monitoring cycle is 1 minute.
- Alarm-triggered Scaling
- Scheduled Scaling
- Replacement of Unhealthy CVM
- Auto Load Balancing
- AS Activity Record
If you want to adjust the business deployment based on CVM metrics, you can customize an alarm policy. When a metric (e.g. CPU utilization, memory utilization, outbound and inbound bandwidth on the private network, outbound and inbound bandwidth on the public network) reaches the threshold due to business load, the policy can automatically increase or decrease the number of CVM instances to deal with the changes in business load flexibly, improving device utilization and reducing costs of deployment and instances. The monitoring cycle is 1 minute.
Scenarios
Business type
E-commerce websites, video websites, online education, etc.
Policy
The requests from the clients reach the application CVMs through CLB. When the access volume changes rapidly, AS can flexibly scale up or down the instances based on the amount of requests.
- Logic-layer CVM Scaling for Flexible Web Services
- High Performance Computing Cluster Deployment
- Request CVM Deployment
Business type
E-commerce websites, video websites, online education, etc.
Policy
The requests from the clients reach the application CVMs through CLB. When the access volume changes rapidly, AS can flexibly scale up or down the instances based on the amount of requests.