Proposed title of this feature request

Setting your own loadbalancers in the install-config.yaml when bootstrapping cluster with IPI

What is the nature and description of the request?

Currently Openshift 4 IPI deploys haproxy + keepalived pods inside the cluster when you deploy openshift4 with IPI. My customer wants to use their own loadbalancer set for deploying the cluster but we don't want to use the UPI install to achieve this.

As the customer describes the current method:

We managed to "move" the api/ingress DNS entries from the VIPs to an external load balancer as a day two-plus operation. But we would rather use the external loadbalancer from the bootstrap phase instead of changing it afterwards. We use this source:

https://cloud.redhat.com/blog/using-integrated-load-balancing-with-on-premises-openshift-4-ipi

From our perspective it would be best if we could define a loadbalancer ip/ hostname that points to our external loadbalancer and the bootstrap VM + master nodes is using this external loadbalancer instead of creating it's own haproxy + keepalived pods.

3. Why does the customer need this? (List the business requirements here)

This issue is causing the customer to have to use "a day 2 plus" operation for every cluster deployed, it would be a much better solution if this could be done right from the install as a setting in the install-config.yaml file. This would ease the customers install while still allowing them to use IPI which we keep asking customers to do.

4. List any affected packages or components.