Now, there are several different ways of deploying Docker containers on the GCP. We could use App Engine Flex, or even simply deploy them on VMs. But choosing either of these options would cause us to lose out on the power of Kubernetes. That is a pretty big loss, because we'd need to manage creating, managing, and scaling our deployments ourselves. It would also tie us to a specific platform—neither Docker on GCP VMs, nor App Engine Flex, would generalize easily to a non-GCP cloud, or to on-premise. Following are the parameters to keep in mind while making a choice between GKE or App Engine Flex.

• Control: The argument is simple here. Containers give a LOT more control and portability. From resource allocation to backend storage provision, everything can be customised with containers which is not the case with App Engine due to it being managed by Google.
• Scaling: Take deployment or scaling, App engine is always faster than GKE cluster specially for the tasks like autoscaling. On the other hand, if HPA (Horizontal Pod Autoscaler) is not employed, scaling on kubernetes can take significantly more time and efforts.
• Cost: Since App Engine Flex still uses managed VMs as its backend, GKE saves a lot of resources (the difference becomes more significant with large deployments) and so is more economic.
• Service: Both App Engine Flex and GKE use GCP’s HTTP(s) load balancing so you can expect similar throughputs apart from the times where