Figure a way to optimize uptime of services (resources handling)

swiss-ai-center / core-engine

The core engine orchestrating all the Swiss AI Center services.

https://docs.swiss-ai-center.ch/reference/core-engine/

GNU Affero General Public License v3.0

6 stars 1 forks source link

Figure a way to optimize uptime of services (resources handling) #184

Open andrptrc opened 7 months ago

andrptrc commented 7 months ago

Actually, each service and component is launched in a Pod that stays up indefinitely. Try testing some alternatives where the pod is launched on execution for example or launched for an execution and then stays up for 1 hour and then shut itself down.

andrptrc commented 7 months ago

Check: https://knative.dev/docs/ https://github.com/apache/openwhisk

ludelafo commented 6 months ago

I think the Kubernetes' StatefulSets can be configured to have autoscaling (minimum 1 instances) and spin up new pods when requests are made as mentioned here: https://stackoverflow.com/a/57732487.

Maybe this can be a way to manage resources as well.