The failed switch has been replaced and the 2 backend servers are back online.
There are still issues with the load balancer which our team is working on and 1/3 of requests will reach a portal server. This has been raised with the hardware vendors team also.