Hi, I don’t know much about prefect but we are using it in our project to run ETL pipelines. Currently we have an algorithm that needs to be run by changing variables A and B, each of which has 3 and 17 different settings respectively for a total of 51 different combinations. Currently we are instantiating some virtual machines and registering N combinations in each machine, for example: VM_1 have A=1 and B=1,2,3,4 ; then this VM will run algorithm A for various B parameters. Then we just register one agent in the prefect and have it run this parameter combination every 1 hour for example and this has been working so far.
But I wonder if there is a smarter way to use prefect, so that instead of having to register each combination on each machine we could make a list of parameters that will be run and prefect would dynamically allocate the resources between the VMs that we have running the prefect agent. It seems that we are using this tool only as a remote cron that allows you to monitor the state of the flows.
For example, use the prefect dashboard or api to pass a list:[ (A=1, B=1), (A=1, B=2), (A=2, B=1), (A=2, B=2),… and the prefect dynamically executes (A=1, B=1) in VM_3, (A=1, B=2) in VM_1 and so on.