I am trying to determine if the streamlined creation via k8s yaml (ing object) is the best way vs a custom bottom approach through other means such as terraform
for one, it seems like health check options are very rigid through the streamlined way (health check override via backendconfig is in the works but I’d assume its not available yet), and 1 ing to 1 gclb ratio seems weird
and id feel more comfortable not tying the lifecycle of the gclb to the ing object, and flexibility to do multi cluster things down the line is nice. (Ingress for Anthos sounds great, but I think “for Anthos” will require some subscription of sorts.)
the custom approach biggest downside is having to add the instance groups as backends imho.
somehow the streamlined way creates these nice unmanaged instance groups: k8s-ig-...