Auto Scaling has both feature constraints and resource quotas. Review these limits before designing your scaling architecture.
Feature limits
Stateless applications only
Apps running on ECS instances or elastic container instances in a scaling group must be stateless and horizontally scalable. Scaling groups add and remove instances automatically, so anything stored locally is lost when an instance is released.
Store persistent data outside the scaling group:
| Data type | Recommended service |
|---|---|
| Session data | A dedicated instance outside the scaling group |
| Application data | ApsaraDB RDS (What is ApsaraDB RDS?) |
| Logs | Simple Log Service (What is Simple Log Service?) |
ApsaraDB for Memcache whitelists
Auto Scaling cannot update ApsaraDB for Memcache whitelists. Add private IP addresses to the whitelist manually. For details, see Configure an IP address whitelist.
Deleted resources are detached automatically
When you delete an associated resource — an ApsaraDB RDS instance, an Application Load Balancer (ALB) instance, a Classic Load Balancer (CLB) instance, or a CLB backend server group — Auto Scaling detaches it from the scaling group automatically.
Automatic scaling is disabled after 30 consecutive days of failures
If scaling operations fail for 30 consecutive days, Auto Scaling disables automatic scaling and alerts you by text message or email. Log in to the Auto Scaling console to diagnose and resolve the issue.
Resource limits
Resource limits fall into two categories: quotas you can request to increase, and fixed limits that cannot be changed.
Adjustable quotas
The following quotas apply per Alibaba Cloud account. To view current values or submit a quota increase request, go to Quota Center. You can also submit a ticket to request a quota increase.
| Resource | Scope |
|---|---|
| Maximum number of scaling groups | Per region |
| Maximum number of scaling configurations | Per scaling group |
| Maximum number of scaling rules | Per scaling group |
| Maximum number of ApsaraDB RDS instances | Per scaling group |
| Maximum number of CLB instances | Per scaling group |
| Maximum number of ALB server groups | Per scaling group |
| Maximum number of Network Load Balancer (NLB) server groups | Per scaling group |
| Maximum number of vServer groups | Per scaling group |
| Maximum number of instances | Per scaling group |
| Maximum number of scheduled tasks | Per region |
Fixed limits
The following limits cannot be changed.
| Resource | Limit |
|---|---|
| Maximum ECS instances or elastic container instances scaled in a single operation | 1,000 |
| Maximum instance types per scaling configuration | 20 |
| Maximum notification rules per scaling group | 6 |
| Maximum lifecycle hooks per scaling group | 10 |