Hi @smm Good Question... I will attempt to answer / clarify.
Pay only for the resources you use
Whether you are using Elastic Cloud or deploying self-managed Elastic software, the core pricing meter for Elastic is based on the underlying resources consumed to run your use case.
No that is not correct, hopefully this helps clarify
What this means when you deploy an Elasticsearch cluster at that point you are consuming Elastic Stack Resources those resources are Elastic Stack component (data nodes, kibana nodes, machine learning nodes etc..).
Elastic Stack Component are Sized / Consumed by RAM Allocation. With that RAM Allocation come appropriate CPU and Persistent Storage. These are based on the predefined HW profiles as mentioned in the article.
To be clear at the point you deploy the resources you are "consuming" them in their entirety and will start paying the full hourly rate whether you actually use the components or not,
You will pay the full hourly rate whether you are actively reading / writing documents or not, whether you are using 0% CPU or 80% CPU or 5% Disk or 80% Disk. You will pay the full hourly rate.
Also To be clear there is "No Pausing" of resources you will pay for the resources whether you are actively using them or they are just sitting idle.
Also there are Data Transfer and Storage Costs as discussed in the article you referenced.
Using the pricing calculator found here
For Example if I deploy
1 x 2GB x 3AZs Hot Nodes = 6GB RAM Hot Total
1 x 1GB x 1 AZ Kibana = 1 GB Kibana
The hourly rate will be $0.2196 per hour.
You will be charged $0.2196 whether you actively use that cluster or not.
Charges are calculated hourly,, rounding up to the nearest hour.
Hope this helps.
BTW we are not all guys