I have a GCP HTTP global load balancer config which is very slow and difficult to manage in the cloud console UI, almost to the point of being unusable. My other configs are very fast and far less complex. The load balancer itself is performing well and is not slow.
When using either Chrome or Firefox it takes up to 30 seconds to get a listing of backends to show after clicking. Looking at the performance tab in the browser dev tools it shows the page framerate dropping to 4FPS and the browser becomes almost totally unresponsive (FF doesn't live through it at all).
The load balancer has:
- 27 backend services (average of 4 backend groups - primarily K8S, using GKE, as well as some unmanaged GCE pools)
- 17 frontends (9 with certs, up to 10 SAN certs with wildcards)
- IPv4 and 6
- 52 host/path rule sets with up to 20 hosts and 10 paths in a single entry (including the default)
Is that an expected behavior for what I've done with it (and so a better approach) or is there something else wrong?