Hey, I'm looking to deploy Flyte to the GCP for data and ML pipelines following the tutorial here: https://docs.flyte.org/en/stable/deployment/gcp/manual.html#create-a-gke-cluster. I'm wondering if someone has experience in setting up a small to medium production cluster, what kind of instance types they used, and some other things to be careful about ...? Thanks!

Hi @Fabio Grätz! Would you mind providing some guidance here? 🙂

Hey 🙂 I’d create one node pool with cpu-only machines, e.g.

n1-standard-2

, on which flyte will run. For the cloud sql database, a

db-n1-standard-2

/

db-custom-2-7680

(same machine but only the latter is accepted when configuring through terraform) is enough. Make sure that the database uses a private service connection as mentioned in the docs. For GPU machines, I recommend you don’t configure node pools with pre-defined types but instead just rely on GKE’s node auto provisioning feature which allows you to create the right-sized nodes on the fly, completely automatically. I recommend you activate GKE’s new image streaming feature which reduces start up times significantly! In case you have any questions while going through the guide, feel free to ping me 🙂