This outlines the different components considered for cost calculations. We consider AWS services that are typically used in a Zero to JupyterHub deployment.
Compute¶
We map AWS services to the compute component as follows:
Amazon Elastic Compute Cloud - Compute
EC2 - Other
Note that the EC2 - Other category includes costs associated with EBS volumes attached to EC2 instances, as well as costs associated with compute for core support nodes.
We therefore subtract the costs for home storage and support from the total compute costs to reflect only the costs associated with compute from user nodes.
Home storage¶
We map AWS services to the home storage component as follows:
AWS Backup
EC2 - Other
We determine costs for the EBS volumes used for home directories from the EC2 - Other service by filtering on the tag 2i2c:volume-purpose=home-nfs.
Support¶
We map AWS services to the support component as follows:
Amazon Elastic Container Service for Kubernetes
EC2 - Other
We determine costs for the following:
root EBS volumes attached to support core nodes by filtering on the tag
2i2c:node-purpose=corehub databases by filtering on the tag
kubernetes.io/created-for/pvc/name= hub-db-dircompute and storage for the Prometheus server, Grafana server and alertmanager by filtering on the tag
kubernetes.io/created-for/pvc/namespace= support
Object storage¶
TBC
Networking¶
TBC