Reliability Functions

Load Balancing

This service works with ELB to balance loads among ECSs created in an AS group.

Health Checks

A health check removes unhealthy ECSs from an AS group. Then, AS adds new ECSs to the AS group so that the number of ECSs is the same as the expected number. There are two types of AS group health checks.

ECS health check: checks the ECS running status. If an ECS is stopped or deleted, it is considered as abnormal. ECS health check is the default health check mode for an AS group. The AS group periodically uses the check result to determine the running status of every ECS in the AS group. If an ECS is unhealthy, AS removes the ECS from the AS group.
ELB health check: determines ECS running status using a load balancing listener. If the AS group uses load balancers, the health check method can also be ELB health check. If you add multiple load balancers to an AS group, an ECS is considered healthy only when all load balancers declare that it is healthy. If any load balancer detects that an ECS is unhealthy, the ECS will be removed from the AS group.

In both ECS health check and ELB health check methods, AS removes unhealthy ECSs from the AS group.

For details, see Auto Scaling Health Check.

Cross-AZ DR

Backend servers of AS can be deployed across AZs. If an AZ is faulty, AS can automatically create ECSs in other AZs to quickly recover services.

If multiple AZs are selected, set Multi-AZ Scaling Policy to Balanced or Sequenced.

Balanced: When scaling out an AS group, the system preferentially distributes ECSs evenly across AZs used by the AS group. If it fails in the target AZ, it automatically selects another AZ based on the sequenced policy.
Sequenced: When scaling out an AS group, the system distributes ECSs to the AZs according to the order in which AZs are specified.