[Base, Advanced, and Enterprise tiers only] You can add or remove instances from an existing system. Adding instances is referred to as scale out. Removing instances is referred to as scale in. Scale out and scale in supports only EBS data storage. A system can always be scaled in after it is scaled out.
- Increases the performance of the system by adding more CPU and memory, and increases the storage bandwidth by decreasing the data volumes managed per node.
- Increases the node count by provisioning additional instances and other resources causing an increase in infrastructure and software costs.
- Consumes additional subnet space. The subnet must have enough IP addresses for the new instances being added to the system.
- Migrates EBS data volumes from existing instances to new instances.
- Decreases the performance of the system by removing CPU and memory, and decreases the storage bandwidth by increasing the data volumes managed per node.
- Decreases the node count by terminating instances that are no longer needed and associated resources causing a decrease in infrastructure and software costs.
- Migrates EBS data volumes from the instances that are removed to the remaining instances.
The database capacity and AMP total count does not change. When scaling out and scaling in, the number of AMPs is redistributed as evenly as possible across the instances.
|To scale out at deployment|
|To scale in at deployment||You cannot scale in at deployment, only after deployment.|
|To scale out after deployment||Scaling Out a System|
|To scale in after deployment||Scaling In a System|