Here's a list of DataDog metrics of Azure Load Balancer that are available to use. It seems like
azure.network_loadbalancers.allocated_snat_ports (count) - Total number of SNAT ports allocated within time period,
azure.network_loadbalancers.snat_connection_count (count) - Total number of new SNAT connections created within time period,
azure.network_loadbalancers.used_snat_ports (count) - Total number of SNAT ports used within time period
are the most releveant.
When SNAT port resources are exhausted, outbound flows fail. You could observe failing outbound connections or are advised by support that you're exhausting SNAT ports.
Simply seeing failed connections does not confirm SNAT exhaustion. Seeing the failed connections was a clue we were having an issue but there is no way to confirm SNAT exhaustion w/o a ticket to Microsoft it turns out.