Summary
Currently, the Batch UI and the billing-related API gives users a cost breakdown of each job/batch based on total runtime. This runtime (right now) necessarily includes cost * time when the VM is waiting for a Docker image to be pulled. A consequence of this is that when comparing performance/cost of otherwise comparable pipelines, the cost of one might be skewed due to nothing more than unusually long docker pull times. This change would allow users to isolate main and running job costs from others.
Motivation
User request, general batch improvement
Detailed Description
No response
Summary
Currently, the Batch UI and the billing-related API gives users a cost breakdown of each job/batch based on total runtime. This runtime (right now) necessarily includes cost * time when the VM is waiting for a Docker image to be pulled. A consequence of this is that when comparing performance/cost of otherwise comparable pipelines, the cost of one might be skewed due to nothing more than unusually long docker pull times. This change would allow users to isolate
mainandrunningjob costs from others.Motivation
User request, general batch improvement
Detailed Description
No response