|Advantages||With AS||Without AS|
No manual operations required
Auto-create and release CVM instances; auto-configure CLB
Tedious manual operations
Resources need to be created and terminated manually, and load balance needs to be configured manually.
No resources wasted
Dynamically create and terminate sub-machines in real time according to loads and policies; supports billing that’s accurate to the second
Lots of waste caused by idle resources
Need to enable extra CVMs for backup
When an exception occurs, an auto-duplicate of a normal instance will instantly replace the abnormal one
Abnormal instances are handled after service interruption is identified, affecting service availability.
Handles common fluctuations smoothly.
Supports time scaling, alarm scaling, APIs and consoles; and relies on resource estimation to an extent.
Handling fluctuations smoothly is difficult.
Resources need to be estimated manually, and instances need to be manually added/removed afterwards.
Add/Remove CVM instantly or periodically
Load-based intelligent scaling
Automatically Add/Remove CVM instances based on cloud monitoring indicators (such as CPU, memory, and network traffic)
Auto registration to CLB whitelist
Auto register new CVMs to CLB, allowing for smooth auto-expansion and shrinkage
Replace failed CVMs
Replace CVMs that cannot be pinged with normal CVMs
CVM instances that are auto-created through AS or manually added are charged the same as pay-by-traffic CVMs.