Typically for ingress (proxy gateway), there would be proxy pods (For e.g IstioGateway, Nginx) and typically a load balancer service in front of it to expose them externally. The proxy pods could be scaled horizontally to handle the load as well as for HA. The Kubernetes service would take care of doing some basic load-balancing of the connections between multiple proxy(gateway) pods.
As load-balancing of Kubernetes services are on connections (L4 level) and not at the request level, How does one manage scaling and HA for these proxy pods(ingress gateway) when the type of requests is HTTP2/GRPC or even for HTTP1(Where Keep-Alive header is set) where clients keep reusing the same connection?
(TLDR: Request-based loadbalancing for the applications behind the Ingress Gateway would be taken care of by the Ingress gateway, but how to take care of loadbalancing the ingress gateway proxies itself for HTTP2/GRPC)
