Currently running riker deploy --scale from scratch does the following:
- Creates a new instances
- Provisions it with docker, nginx, and any other dependencies
- Creates AMI (A)
- Terminates instance
- Launches instance from AMI (A)
- Deploys app to instance
- Creates AMI (B)
- Terminates instance
- Launches instance from AMI (B) (only if
--scale isn't provided), or adds AMI (B) to an auto-scaling group to launch instance.
Can be compressed by not terminating the instance in #4, therefore skipping #5. If --scale is not provided, #8 and #9 can be skipped too. If --scale is provided, then that instance can be added directly to the auto-scaling group.
This will greatly improve deployment time, as AMI creation and instance termination take quite a bit of time.
Currently running
riker deploy --scalefrom scratch does the following:--scaleisn't provided), or adds AMI (B) to an auto-scaling group to launch instance.Can be compressed by not terminating the instance in
#4, therefore skipping#5. If--scaleis not provided,#8and#9can be skipped too. If--scaleis provided, then that instance can be added directly to the auto-scaling group.This will greatly improve deployment time, as AMI creation and instance termination take quite a bit of time.