So basically from the course I understood the following:
1. you create a pod with a request of X resource – this will be allocated (reserved) from the VM/servers
2. you add a HPA that will trigger autoscalling when X/2 of that resource request is hit.
3. deployment autoscales -> the average usage of resources per pod goes down ..BUT
4. new pod -> new request -> more blocked resources from the VMs
5. basically having a HPA that autoscales at X/2 of the requested resources will trigger the following:
– you will basically really use in computing only 1/2 of the requested resource overall for you microservice because each time this goes over X/2 , a new pod will be spawned for getting everything back to lower than X/2 . Isn’t this meaning a waste of resources? The more you spawn the more actual resources you WASTE and obviously use:)