Open balewski opened 1 year ago
Thank you for the request, this would be a nice feature addition to the cost tracking currently offered in the Braket SDK. Could you help us understand some more context around your use case? Are you wanting the cost tracking for informational purposes? Are you trying to use the cost for control flow in your code? Or something else? Have you tried the AWS wide tools offered by Cost Explorer and AWS Budgets?
It is for informal cost tracking purposes. I have not tried yet the tools you mentioned. Can you point me to a tutorial? I need nothing fancy, just a basic request for the cost of the list of jobs.
đŸ‘‹ I'm on the product team for Braket - cost management is a theme we are planning to expand on. In the interim, these two links shed light on AWS Cost Explore and Budgets. They provide higher level information for the Braket service, but your point on asking for granular cost information is valid.
Is your feature request related to a problem? Please describe. I'd like to be able to track acquired cost per task, after task has been executed.
When I submit a task to Aquila it may take a week until it gets executed. You provide the construct, which requires the Tracker() context to be active for the duration of the job.
It takes a week+ for Aquila job to get executed and it is unrealistic to expect a program on my laptop maintains AWS connection for that long. If I want to place 20 jobs in the queue I'd need to keep 20 copies alive for that long.
Describe the solution you'd like Given a list of ARNs I'd like to construct a CSV table contacting information reported by
using a Python script executed on a terminal.
Describe alternatives you've considered Open 20 notebook for 20 submitted jobs and not close my laptop for a week+, until all complete.